CN112100528A - Method, device, equipment and medium for training search result scoring model - Google Patents
- ️Fri Dec 18 2020
CN112100528A - Method, device, equipment and medium for training search result scoring model - Google Patents
Method, device, equipment and medium for training search result scoring model Download PDFInfo
-
Publication number
- CN112100528A CN112100528A CN202010941415.4A CN202010941415A CN112100528A CN 112100528 A CN112100528 A CN 112100528A CN 202010941415 A CN202010941415 A CN 202010941415A CN 112100528 A CN112100528 A CN 112100528A Authority
- CN
- China Prior art keywords
- search result
- search
- pair
- terminal
- page Prior art date
- 2020-09-09 Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9538—Presentation of query results
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/30—Authentication, i.e. establishing the identity or authorisation of security principals
- G06F21/31—User authentication
- G06F21/32—User authentication using biometric data, e.g. fingerprints, iris scans or voiceprints
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computer Security & Cryptography (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Computer Hardware Design (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application discloses a method, a device, equipment and a medium for training a search result scoring model, and belongs to the technical field of computers. The method comprises the following steps: receiving a search request carrying first search information sent by a first terminal, and sending a plurality of search results corresponding to the first search information to the first terminal; acquiring operation information of each search result after the first terminal displays the plurality of search results, and determining an operation class corresponding to the operation information of each search result; determining a reference score corresponding to each search result based on a pre-stored reference score corresponding to each operation class and an operation class corresponding to each search result; and training a search result scoring model based on the plurality of search results and the reference score corresponding to each search result. By the aid of the method and the device, training effect of training of the search result scoring model can be improved.
Description
Technical Field
The present application relates to the field of computer technologies, and in particular, to a method, an apparatus, a device, and a medium for training a search result scoring model.
Background
With the development of computer technology, most people can obtain relevant answers by searching when problems which cannot be solved in daily life of people occur. When the search processing is performed, the server obtains the scores of the search results through the search result scoring model, and then ranks the search results according to the scores of the search results. The above-mentioned search result scoring model may be trained as follows:
the method comprises the steps of receiving a search request which is sent by a terminal and carries target search information, sending a plurality of search results corresponding to the target search information to the terminal, obtaining historical click times corresponding to each search result, determining a benchmark score corresponding to each search result based on the historical click times, and training a search result scoring model based on the benchmark score corresponding to each search result and the plurality of search results.
In the process of implementing the present application, the inventor finds that the prior art has at least the following problems:
the benchmark score in the scheme is obtained based on the historical click times, but the benchmark score also comprises a plurality of operations when the benchmark score is actually used by a user, so that the granularity of the search result scoring model in the scheme is coarse, and the judgment accuracy of the search result is poor.
Disclosure of Invention
The embodiment of the application provides a method, a device, equipment and a medium for training a search result scoring model, and can solve the problems of coarse scoring granularity of the search result scoring model and poor judging accuracy of a search result. The technical scheme is as follows:
in one aspect, a method of training a search result scoring model is provided, the method comprising:
receiving a search request carrying first search information sent by a first terminal, and sending a plurality of search results corresponding to the first search information to the first terminal;
acquiring operation information of the first terminal on each search result after the plurality of search results are displayed, and determining an operation class corresponding to the operation information of each search result, wherein the operation information is used for indicating whether the first terminal triggers the search result after the search result is displayed for a first preset time or before a target function is triggered, and if the search result is triggered, the first terminal displays the time of a page corresponding to the search result and whether the target function is triggered through the page;
determining a reference score corresponding to each search result based on a pre-stored reference score corresponding to each operation class and an operation class corresponding to each search result;
and training a search result scoring model based on the plurality of search results and the reference score corresponding to each search result.
Optionally, the operation class includes at least one of:
before reaching a first preset time or triggering a target function, the first terminal does not trigger the search result;
before reaching a first preset time or triggering a target function, the first terminal triggers the search result to display a corresponding page and closes the page, and the target function is not triggered through the page, wherein the display time of the page is less than a second preset time;
before reaching a first preset time or triggering a target function, the first terminal triggers the search result to display a corresponding page and close the page, and the target function is not triggered by the page, wherein the display time of the page is longer than a second preset time and shorter than a third preset time;
before the first preset duration is reached, the first terminal triggers the search result to display a corresponding page, and triggers the target function through the page;
the first preset time length is longer than a third preset time length, and the third preset time length is longer than the second preset time length.
Optionally, the training a search result scoring model based on the plurality of search results and the reference score corresponding to each search result includes:
pairing the plurality of search results to obtain at least one search result pair, wherein two search results in each search result pair correspond to different operation classes;
respectively inputting each search result into a search result scoring model to obtain an actual score corresponding to each search result, and determining the actual difference of each search result pair based on the actual scores corresponding to two search results in each search result pair;
determining the benchmark difference degree of each search result pair based on the benchmark scores corresponding to the two search results in each search result pair;
training the search result scoring model based on the actual difference and the reference difference of each search result pair.
Optionally, the determining the actual difference degree of each search result pair based on the actual scores corresponding to two search results in each search result pair includes:
and determining the actual difference degree of each search result pair based on the magnitude relation of the actual scores corresponding to the two search results in each search result pair.
Optionally, the determining the reference difference degree of each search result pair based on the reference scores corresponding to two search results in each search result pair includes:
and determining the benchmark difference degree of each search result pair based on the magnitude relation of the benchmark scores corresponding to the two search results of each search result pair.
Optionally, the two search results in each search result pair are arranged according to a specified order, and the determining the reference difference degree of each search result pair based on the magnitude relation between the reference scores corresponding to the two search results of each search result pair includes:
for each search result pair, if the reference score corresponding to the search result ranked in the front of the search result pair is greater than the reference score corresponding to the search result ranked in the back of the search result pair, determining that the reference difference degree corresponding to the search result pair is 1, and if the reference score corresponding to the search result ranked in the front of the search result pair is less than the reference score corresponding to the search result ranked in the back of the search result pair, determining that the reference difference degree corresponding to the search result pair is 0.
Optionally, the determining the actual difference degree of each search result pair based on the magnitude relationship between the actual scores corresponding to the two search results in each search result pair includes:
for each search result pair, by formula
Determining the actual degree of dissimilarity g (item) for each search result pairi),f(itemj) Wherein, the f (item)i) The actual score of the top ranked search result of the search result pair, f (item)j) The actual score of the top ranked search result of the search result pair is assigned.
Optionally, the training the search result scoring model based on the actual difference and the reference difference of each search result pair includes:
and for each search result pair, inputting the actual difference and the reference difference corresponding to the search result pair into a loss function to obtain an adjustment value of a parameter in the search result scoring model, and updating the search result scoring model based on the adjustment value of the parameter.
Optionally, after the training of the search result scoring model based on the plurality of search results and the reference score corresponding to each search result, the method further includes:
receiving a search request carrying second search information sent by a second terminal;
obtaining a plurality of search results corresponding to the second search information;
determining a score corresponding to each search result based on the second search information, each search result and the trained search result scoring model;
ranking the search results based on the score corresponding to each search result to obtain a search result list;
and sending the search result list to the second terminal.
In another aspect, an apparatus for training a search result scoring model is provided, the apparatus comprising:
the device comprises a sending module, a searching module and a searching module, wherein the sending module is used for receiving a searching request which is sent by a first terminal and carries first searching information and sending a plurality of searching results corresponding to the first searching information to the first terminal;
an obtaining module, configured to obtain operation information of the first terminal for each search result after the plurality of search results are displayed, and determine an operation class corresponding to the operation information of each search result, where the operation information is used to indicate whether the first terminal triggers the search result after the search result is displayed for a first preset time or before a target function is triggered, and if the search result is triggered, the first terminal triggers a display time of a page corresponding to the search result and whether the target function is triggered by the page;
the determining module is used for determining the reference score corresponding to each search result based on the pre-stored reference score corresponding to each operation class and the operation class corresponding to each search result;
and the training module is used for training the search result scoring model based on the plurality of search results and the reference score corresponding to each search result.
Optionally, the operation class includes at least one of:
before reaching a first preset time or triggering a target function, the first terminal does not trigger the search result;
before reaching a first preset time or triggering a target function, the first terminal triggers the search result to display a corresponding page and closes the page, and the target function is not triggered through the page, wherein the display time of the page is less than a second preset time;
before reaching a first preset time or triggering a target function, the first terminal triggers the search result to display a corresponding page and close the page, and the target function is not triggered by the page, wherein the display time of the page is longer than a second preset time and shorter than a third preset time;
before the first preset duration is reached, the first terminal triggers the search result to display a corresponding page, and triggers the target function through the page;
the first preset time length is longer than a third preset time length, and the third preset time length is longer than the second preset time length.
Optionally, the training module is configured to:
pairing the plurality of search results to obtain at least one search result pair, wherein two search results in each search result pair correspond to different operation classes;
respectively inputting each search result into a search result scoring model to obtain an actual score corresponding to each search result, and determining the actual difference of each search result pair based on the actual scores corresponding to two search results in each search result pair;
determining the benchmark difference degree of each search result pair based on the benchmark scores corresponding to the two search results in each search result pair;
training the search result scoring model based on the actual difference and the reference difference of each search result pair.
Optionally, the training module is configured to:
and determining the actual difference degree of each search result pair based on the magnitude relation of the actual scores corresponding to the two search results in each search result pair.
Optionally, the training module is configured to:
and determining the benchmark difference degree of each search result pair based on the magnitude relation of the benchmark scores corresponding to the two search results of each search result pair.
Optionally, two search results in each search result pair are arranged according to a specified order, and the training module is configured to:
for each search result pair, if the reference score corresponding to the search result ranked in the front of the search result pair is greater than the reference score corresponding to the search result ranked in the back of the search result pair, determining that the reference difference degree corresponding to the search result pair is 1, and if the reference score corresponding to the search result ranked in the front of the search result pair is less than the reference score corresponding to the search result ranked in the back of the search result pair, determining that the reference difference degree corresponding to the search result pair is 0.
Optionally, the training module is configured to:
for each search result pair, by formula
Determining the actual degree of dissimilarity g (item) for each search result pairi),f(itemj) Wherein, the f (item)i) The actual score of the top ranked search result of the search result pair, f (item)j) The actual score of the top ranked search result of the search result pair is assigned.
Optionally, the training module is configured to:
and for each search result pair, inputting the actual difference and the reference difference corresponding to the search result pair into a loss function to obtain an adjustment value of a parameter in the search result scoring model, and updating the search result scoring model based on the adjustment value of the parameter.
Optionally, the apparatus further includes a search module, where the search module is configured to:
receiving a search request carrying second search information sent by a second terminal;
obtaining a plurality of search results corresponding to the second search information;
determining a score corresponding to each search result based on the second search information, each search result and the trained search result scoring model;
ranking the search results based on the score corresponding to each search result to obtain a search result list;
and sending the search result list to the second terminal.
In yet another aspect, a computer device is provided, which includes a processor and a memory, the memory having instructions stored therein, and the processor executing the instructions causes the computer device to implement the method of training a search result scoring model.
In yet another aspect, a computer-readable storage medium is provided, which stores instructions that, when executed by a computer device, cause the computer device to implement the method for training a search result scoring model.
The technical scheme provided by the embodiment of the application has the following beneficial effects:
according to the method and the device, the operation class corresponding to the operation information of each search result is determined by obtaining the operation information of the first terminal after the first terminal displays the search results, then the reference score corresponding to each search result is determined based on the pre-stored reference score corresponding to each operation class and the operation class corresponding to each search result, and further the reference score corresponding to each search result is based on the search results and the search results, the search result scoring model is trained.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIG. 1 is a schematic diagram of an implementation environment provided by an embodiment of the present application;
FIG. 2 is a flowchart of a method for training a search result scoring model according to an embodiment of the present disclosure;
FIG. 3 is a flow chart of a method for using a search result scoring model according to an embodiment of the present application;
fig. 4 is a schematic structural diagram of an apparatus for training a search result scoring model according to an embodiment of the present disclosure;
fig. 5 is a schematic structural diagram of a terminal provided in an embodiment of the present application;
fig. 6 is a schematic structural diagram of a server according to an embodiment of the present application.
Detailed Description
To make the objects, technical solutions and advantages of the present application more clear, embodiments of the present application will be described in further detail below with reference to the accompanying drawings.
The embodiment of the application provides a method for training a search result scoring model, which can be realized by a first terminal and a server together. The first terminal can be a mobile phone, a desktop computer, a tablet computer, an intelligent wearable device and the like. The first terminal may have a function of receiving data and a function of sending data, and an application program that can perform a search, such as a takeaway application program, may be installed on the first terminal. Here, the first terminal is a terminal used for training, and the user or the developer using the first terminal may be used. The server may be a background server of the takeout application program, the server may be a single server or a server group, if the server is a single server, the server may be responsible for all processing that needs to be performed by the server in the following scheme, if the server group is a server group, different servers in the server group may be respectively responsible for different processing in the following scheme, and a specific processing allocation condition may be arbitrarily set by a technician according to an actual requirement, which is not described herein again.
As shown in fig. 1, in daily life, when using a search function of a takeout application, a server may receive a search request carrying first search information, where the first search information at least includes a keyword, and the server may obtain a plurality of search results corresponding to the keyword, and then score the search results to determine an arrangement order of the search results.
Fig. 2 is a flowchart of a method for training a search result scoring model according to an embodiment of the present disclosure. Referring to fig. 2, the process includes:
201, receiving a search request carrying first search information sent by a first terminal, and sending a plurality of search results corresponding to the first search information to the first terminal.
The first search information may include a keyword, a current location, and the like.
In an implementation, a user (or a developer, which will not be described in detail hereinafter) may use the first terminal to open the takeaway application and enter a keyword, such as a fried chicken, in a search box of the takeaway application. Then, a search control is triggered. Further, the first terminal may obtain a current location, and then generate a search request based on the current location and the keyword. Then, the first terminal can send the search request to the server, and the server obtains a plurality of corresponding search results according to the keyword carried in the search request and the current position. Meanwhile, the server can also score the plurality of search results, sort the plurality of search results according to the obtained scores to obtain sorting information of the plurality of search results, and then send the sorting information and the plurality of search results to the first terminal.
202, obtaining operation information of each search result after the first terminal displays the plurality of search results, and determining an operation class corresponding to the operation information of each search result.
The operation information is used for indicating whether the first terminal triggers the search result after the search result is displayed for a first preset time or before the target function is triggered, and if the search result is triggered, the first terminal displays the time of the page corresponding to the search result and whether the target function is triggered through the page.
In implementation, after the first terminal receives the plurality of search results and the ranking information corresponding to the first search information, the first terminal may display the plurality of search results on the screen according to the ranking information, and then the user may browse each search result, and may also trigger a control corresponding to each search result, and then the user may browse all contents corresponding to the triggered search result. When the user browses the search result displayed by the first terminal or all contents corresponding to the search result, the terminal may record operation information of the user on each search result, where the operation information may include a display duration of a page and a trigger of the user. And then, when the first preset time length is reached or the target function is triggered, the first terminal sends the operation information of each search result to the server. The server may store the operation information of each search result after receiving the operation information of each search result.
Further, the server may determine the operation class corresponding to the operation information of each search result by:
and if the first terminal does not trigger the search result before the first preset time length is reached or the target function is triggered, determining the search result corresponding to the operation information as a first operation class.
The first preset time length is used for forcibly ending the time length for recording the operation information when the display time length of the page is longer than the first preset time length, and also forcibly ending the recording of the operation information after the target function is triggered.
In implementation, if no trigger of the search result is recorded in the operation information, the search result corresponding to the operation information is determined as the first operation class.
For example, when the user uses the takeaway application program, after the user finds that the search keyword is 'chicken frying', the displayed search results do not meet the requirements of the user, and the user cannot trigger any control corresponding to the search result until the first preset time is reached. Correspondingly, when the operation information is generated, the user does not trigger the control corresponding to any search result, and the display duration of the page is only recorded in the operation information. Further, it may be determined from the operation information of the search result that it belongs to the first operation class.
And if the first terminal triggers the search result to display the corresponding page and closes the page before the first preset time or the target function is triggered, and the search result corresponding to the operation information is determined as a second operation class if the target function is not triggered through the page.
The first preset time is used for forcibly ending the time for recording the operation information when the display time of the page is longer than the first preset time, the recording of the operation information can be forcibly ended after the target function is triggered, the display time of the page is shorter than the second preset time, the second preset time is used for judging whether the user carries out misoperation, according to the experience of developers, the user can quit the page within the second preset time if the user enters the page corresponding to one search result through misoperation, namely, if the display time of the page is shorter than the second preset time, the user is considered to carry out misoperation.
In implementation, if the operation information records that a user triggers a search result, and a control for closing a page is triggered after the corresponding page is displayed on the first terminal, and the operation information does not record a trigger target function in the page, if the display duration of the page is less than a second preset duration, the search result corresponding to the operation information is determined as a second operation class.
For example, when the user uses the takeout application program, the first terminal can display a plurality of search results after the user finds the search keyword 'chicken frying', the user triggers the control for closing the page when the user triggers the control corresponding to the search results, and then the user triggers the control for closing the page when the user finds that the dishes displayed in the page do not meet the requirements of the user, and then the user can perform the operation for a plurality of times, wherein the display duration of each page does not exceed the second preset duration until the first preset duration is reached. Correspondingly, when the operation information is generated, because the user triggers the control corresponding to the at least one search result and the control for closing the page in the at least one page, the display duration of each page, the event that the control corresponding to the search result is triggered, and the event that the control for closing the page in the page is triggered are recorded in the operation information. Further, it may be determined from the operation information of the search result that it belongs to the second operation class.
And if the first terminal triggers the search result to display the corresponding page and closes the page before the first preset time or the target function is triggered, and the search result corresponding to the operation information is determined as a third operation class if the target function is not triggered through the page.
The first preset time is used for forcibly ending the time for recording the operation information when the display time of the page is longer than the first preset time, the recording of the operation information can be forcibly ended after the target function is triggered, the display time of the page is longer than the second preset time and shorter than the third preset time, the second preset time and the third preset time are used for judging whether the user carries out misoperation, and according to the experience of developers, if the user does not enter the page corresponding to one search result by misoperation, the user can quit the page from the second preset time to the third preset time.
In implementation, if the operation information records that a user triggers a search result, and a control for closing a page is triggered after the first terminal displays a corresponding page, and the operation information does not record a trigger target function in the page, if the display duration of the page is longer than a second preset duration and shorter than a third preset duration, the search result corresponding to the operation information is determined as a third operation class.
For example, when the user uses the takeout application program, the first terminal can display a plurality of search results after the user finds the search keyword 'chicken frying', the user selects a plurality of dishes in the page after the user triggers the control corresponding to the search results, but the user triggers the control for closing the page when the user finds that the dishes do not meet the requirements of the user very much, then the user can perform the operation for many times, and the display duration of the page of each page is longer than the third preset duration within the second preset duration until the first preset duration is reached. Correspondingly, when the operation information is generated, because the user triggers the control corresponding to the at least one search result and the control for closing the page in the at least one page, the display duration of each page, the event that the control corresponding to the search result is triggered, and the event that the control for closing the page in the page is triggered are recorded in the operation information. Further, it may be determined from the operation information of the search result that it belongs to the third operation class.
And if the first terminal triggers the search result to display the corresponding page before the first preset time length is reached, and the target function is triggered through the page, determining the search result corresponding to the operation information as a fourth operation class.
The first preset time length is used for forcibly ending the time length for recording the operation information when the display time length of the page is longer than the first preset time length, and also forcibly ending the recording of the operation information after the target function is triggered.
In implementation, if an event that the user triggers the search result is recorded in the operation information and an event that the control of the target function is triggered through the page is recorded in the operation information, the search result corresponding to the operation information is determined as the fourth operation class.
For example, when the user is a takeaway application, after the user finds the search keyword "fry chicken", the first terminal may display a plurality of search results, and after the user triggers the control corresponding to the search results, the user selects several dishes in the page and triggers the "submit an order" control. Correspondingly, when the operation information is generated, since the user triggers the control corresponding to at least one search result and the control for "submitting an order", the display duration of each page, the event that the control corresponding to the search result is triggered, and the event that the control for "submitting an order" is triggered are recorded in the operation information. Further, it may be determined from the operation information of the search result that it belongs to the fourth operation class.
203, determining a reference score corresponding to each search result based on the pre-stored reference score corresponding to each operation class and the operation class corresponding to each search result.
In implementation, after determining the operation class corresponding to the operation information of each search result, the server may determine the reference score corresponding to each search result according to the pre-stored reference score corresponding to each operation class and the pre-stored operation class corresponding to each search result.
For example, if the first operation class, the second operation class, the third operation class, and the fourth operation class correspond to reference scores of 1, 2, 3, and 4, respectively, the server determines that the reference score of the search result corresponding to the first operation class is 1, the reference score of the search result corresponding to the second operation class is 2, the reference score of the search result corresponding to the third operation class is 3, and the reference score of the search result corresponding to the fourth operation class is 4.
And step 204, training a search result scoring model based on the plurality of search results and the reference score corresponding to each search result.
In implementation, after determining the benchmark score corresponding to each search result, the following processing may be performed:
first, a plurality of search results are paired to obtain at least one search result pair.
Wherein two search results in each search result pair correspond to different operation classes.
In implementation, after obtaining the classification corresponding to each search result, the server may pair two search results of different classifications to obtain at least one search result pair. The two search results in each search result pair can be individually itemiAnd itemjTo indicate.
Secondly, inputting each search result into the search result scoring model respectively to obtain an actual score corresponding to each search result, and determining the actual difference degree of each search result pair based on the actual scores corresponding to the two search results in each search result pair.
In implementation, after obtaining at least one search result pair, the server may input each search result into the search result scoring model, respectively, to obtain an actual score corresponding to each search result. The server may then determine the actual degree of difference for each pair of search results based on the magnitude relationship of the actual scores for the two search results in each pair.
For example, in obtaining search result pairs (item)i，itemj) Then, the server stores the itemiAnd itemjRespectively inputting the scoring models of the search results to obtain actual scores f (item) corresponding to the two search resultsi) And f (item)j) The server may then formulate
The actual degree of difference for each search result pair is determined.
Wherein, f (item)i) Actual score, f (item), for top ranked search result of the search result pairj) The actual score of the top ranked search result of the search result pair is assigned.
Secondly, determining the benchmark difference degree of each search result pair based on the benchmark scores corresponding to the two search results in each search result pair.
In implementation, the reference scores corresponding to two search results in each search result pair based on the processing in
step203 are further determined, that is, for each search result pair, if the reference score corresponding to the search result ranked in the search result pair before is greater than the reference score corresponding to the search result ranked after, the reference score corresponding to the search result pair before is determined to be 1, and if the reference score corresponding to the search result ranked in the search result pair before is less than the reference score corresponding to the search result ranked after, the reference score corresponding to the search result pair before is determined to be 0.
For example, itemiHas a base score of 1, itemjHas a base score of 3, itemjItem for the top ranked search result of the search result pairjAnd determining that the reference difference degree corresponding to the search result pair is 0 if the search result pair is the ranked search result in the search result pair.
Then, based on the actual difference and the reference difference of each search result pair, a search result scoring model is trained.
In implementation, after obtaining the actual difference and the reference difference of each search result pair, the server may input the actual difference and the reference difference corresponding to the search result pair into the loss function to obtain an adjustment value of a parameter in the search result scoring model, and update the search result scoring model based on the adjustment value of the parameter.
For example, the actual degree of difference and the reference degree of difference are input to Cost (z, g) — zlogg- (1-z) log (1-g), where z represents the reference degree of difference and g represents the actual degree of differenceThe degree of difference, and thus Cost (z, g), is then based on
And (5) carrying out gradient reduction on Cost (z, g) to obtain an adjustment value of a parameter in the search result scoring model, wherein w in the formula is the parameter of the search result scoring model, and the server can update the search result scoring model based on the adjustment value of the parameter.
As shown in fig. 3, after training the search result scoring model, the following processes may be performed:
301, receiving a search request carrying second search information sent by a second terminal.
In an implementation, after the training of the search result scoring model is completed, the user may use the second terminal to open the takeaway application and enter a keyword, such as a fried chicken, in the search box of the takeaway application. Then, a search control is triggered. Further, the second terminal may obtain the current location, and then generate a search request based on the current location and the keyword. Then, the first terminal may transmit the search request to the server. And the server may receive a search request carrying the second search information (that is, carrying the current location and the keyword) sent by the second terminal.
302, determining a score corresponding to each search result based on the second search information, each search result and the trained search result scoring model.
In implementation, after receiving a search request carrying second search information (that is, carrying the current location and the keyword), the server may obtain a plurality of corresponding search results according to the keyword carried in the search request and the current location. And the server can input each search result and the second search information into the trained search result scoring model to determine the score corresponding to each search result.
303, ranking the search results based on the score corresponding to each search result to obtain a search result list.
In implementation, after determining the score corresponding to each search result, the server may rank the plurality of search results based on the size of the score, and further obtain the search result list.
304, sending the search result list to the second terminal.
In an implementation, after obtaining the search result list, the server may send the search result list to the second terminal. Further, the second terminal may display the search result list for the user to browse.
According to the method and the device, the operation class corresponding to the operation information of each search result is determined by obtaining the operation information of the first terminal after the first terminal displays the search results, then the reference score corresponding to each search result is determined based on the pre-stored reference score corresponding to each operation class and the operation class corresponding to each search result, and further the reference score corresponding to each search result is based on the search results and the search results, the search result scoring model is trained.
All the above optional technical solutions may be combined arbitrarily to form the optional embodiments of the present disclosure, and are not described herein again.
An apparatus for audio transmission feedback according to an embodiment of the present application is a server according to the foregoing embodiment, and as shown in fig. 4, the apparatus includes:
a sending
module410, configured to receive a search request carrying first search information sent by a first terminal, and send a plurality of search results corresponding to the first search information to the first terminal;
an obtaining
module420, configured to obtain operation information of each search result obtained after the first terminal displays the plurality of search results, and determine an operation class corresponding to the operation information of each search result, where the operation information is used to indicate whether the first terminal triggers the search result after the first terminal displays the search result and before the first terminal reaches a first preset time or triggers a target function, and if the first terminal triggers the search result, whether the first terminal triggers the target function through the display time of a page corresponding to the search result and the page;
a determining
module430, configured to determine a reference score corresponding to each search result based on a pre-stored reference score corresponding to each operation class and an operation class corresponding to each search result;
a
training module440, configured to train a search result scoring model based on the plurality of search results and the reference score corresponding to each search result.
Optionally, the operation class includes at least one of:
before reaching a first preset time or triggering a target function, the first terminal does not trigger the search result;
before reaching a first preset time or triggering a target function, the first terminal triggers the search result to display a corresponding page and closes the page, and the target function is not triggered through the page, wherein the display time of the page is less than a second preset time;
before reaching a first preset time or triggering a target function, the first terminal triggers the search result to display a corresponding page and close the page, and the target function is not triggered by the page, wherein the display time of the page is longer than a second preset time and shorter than a third preset time;
before the first preset duration is reached, the first terminal triggers the search result to display a corresponding page, and triggers the target function through the page;
the first preset time length is longer than a third preset time length, and the third preset time length is longer than the second preset time length.
Optionally, the
training module440 is configured to:
pairing the plurality of search results to obtain at least one search result pair, wherein two search results in each search result pair correspond to different operation classes;
respectively inputting each search result into a search result scoring model to obtain an actual score corresponding to each search result, and determining the actual difference of each search result pair based on the actual scores corresponding to two search results in each search result pair;
determining the benchmark difference degree of each search result pair based on the benchmark scores corresponding to the two search results in each search result pair;
training the search result scoring model based on the actual difference and the reference difference of each search result pair.
Optionally, the
training module440 is configured to:
and determining the actual difference degree of each search result pair based on the magnitude relation of the actual scores corresponding to the two search results in each search result pair.
Optionally, the
training module440 is configured to:
and determining the benchmark difference degree of each search result pair based on the magnitude relation of the benchmark scores corresponding to the two search results of each search result pair.
Optionally, the two search results in each search result pair are arranged according to a specified order, and the
training module440 is configured to:
for each search result pair, if the reference score corresponding to the search result ranked in the front of the search result pair is greater than the reference score corresponding to the search result ranked in the back of the search result pair, determining that the reference difference degree corresponding to the search result pair is 1, and if the reference score corresponding to the search result ranked in the front of the search result pair is less than the reference score corresponding to the search result ranked in the back of the search result pair, determining that the reference difference degree corresponding to the search result pair is 0.
Optionally, the
training module440 is configured to:
for each search result pair, by formula
Determining each search result pairActual degree of difference g (item)i),f(itemj) Wherein, the f (item)i) The actual score of the top ranked search result of the search result pair, f (item)j) The actual score of the top ranked search result of the search result pair is assigned.
Optionally, the
training module440 is configured to:
and for each search result pair, inputting the actual difference and the reference difference corresponding to the search result pair into a loss function to obtain an adjustment value of a parameter in the search result scoring model, and updating the search result scoring model based on the adjustment value of the parameter.
Optionally, the apparatus further includes a search module, where the search module is configured to:
receiving a search request carrying second search information sent by a second terminal;
obtaining a plurality of search results corresponding to the second search information;
determining a score corresponding to each search result based on the second search information, each search result and the trained search result scoring model;
ranking the search results based on the score corresponding to each search result to obtain a search result list;
and sending the search result list to the second terminal.
According to the method and the device, the operation class corresponding to the operation information of each search result is determined by obtaining the operation information of the first terminal after the first terminal displays the search results, then the reference score corresponding to each search result is determined based on the pre-stored reference score corresponding to each operation class and the operation class corresponding to each search result, and further the reference score corresponding to each search result is based on the search results and the search results, the search result scoring model is trained.
It should be noted that: in the apparatus for training a score model of a search result provided in the foregoing embodiment, when the score model of the search result is trained, only the division of the function modules is illustrated, and in practical applications, the function distribution may be completed by different function modules according to needs, that is, the internal structure of the device may be divided into different function modules, so as to complete all or part of the functions described above. In addition, the embodiments of the method for training the search result scoring model provided in the above embodiments belong to the same concept, and specific implementation processes thereof are described in the embodiments of the method for training the search result scoring model, and are not described herein again.
Fig. 5 shows a block diagram of a terminal 500 according to an exemplary embodiment of the present application. The terminal may be the first terminal and the second terminal in the above embodiments, and the terminal 500 may be: a smart phone, a tablet computer, an MP3 player (Moving Picture Experts Group Audio Layer III, motion video Experts compression standard Audio Layer 3), an MP4 player (Moving Picture Experts Group Audio Layer IV, motion video Experts compression standard Audio Layer 4), a notebook computer, or a desktop computer.
Terminal500 may also be referred to by other names such as user equipment, portable terminal, laptop terminal, desktop terminal, and the like.
In general, the terminal 500 includes: a
processor501 and a
memory502.
The
processor501 may include one or more processing cores, such as a 4-core processor, an 8-core processor, and so on. The
processor501 may be implemented in at least one hardware form of a DSP (Digital Signal Processing), an FPGA (Field-Programmable Gate Array), and a PLA (Programmable Logic Array). The
processor501 may also include a main processor and a coprocessor, where the main processor is a processor for Processing data in an awake state, and is also called a Central Processing Unit (CPU); a coprocessor is a low power processor for processing data in a standby state. In some embodiments, the
processor501 may be integrated with a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content required to be displayed on the display screen. In some embodiments,
processor501 may also include an AI (Artificial Intelligence) processor for processing computational operations related to machine learning.
502 may include one or more computer-readable storage media, which may be non-transitory.
Memory502 may also include high-speed random access memory, as well as non-volatile memory, such as one or more magnetic disk storage devices, flash memory storage devices. In some embodiments, a non-transitory computer readable storage medium in
memory502 is used to store at least one instruction for execution by
processor501 to implement the method of training a search result scoring model provided by method embodiments herein.
In some embodiments, the terminal 500 may further optionally include: a
peripheral interface503 and at least one peripheral. The
processor501,
memory502 and
peripheral interface503 may be connected by a bus or signal lines. Each peripheral may be connected to the
peripheral interface503 by a bus, signal line, or circuit board. Specifically, the peripheral device includes: at least one of
radio frequency circuitry504,
touch screen display505,
camera506,
audio circuitry507,
positioning components508, and
power supply509.
The
peripheral interface503 may be used to connect at least one peripheral related to I/O (Input/Output) to the
processor501 and the
memory502. In some embodiments, the
processor501,
memory502, and
peripheral interface503 are integrated on the same chip or circuit board; in some other embodiments, any one or two of the
processor501, the
memory502, and the
peripheral interface503 may be implemented on a separate chip or circuit board, which is not limited in this embodiment.
The
Radio Frequency circuit504 is used for receiving and transmitting RF (Radio Frequency) signals, also called electromagnetic signals. The
radio frequency circuitry504 communicates with communication networks and other communication devices via electromagnetic signals. The
rf circuit504 converts an electrical signal into an electromagnetic signal to transmit, or converts a received electromagnetic signal into an electrical signal. Optionally, the
radio frequency circuit504 includes: an antenna system, an RF transceiver, one or more amplifiers, a tuner, an oscillator, a digital signal processor, a codec chipset, a subscriber identity module card, and so forth. The
radio frequency circuitry504 may communicate with other terminals via at least one wireless communication protocol. The wireless communication protocols include, but are not limited to: metropolitan area networks, various generation mobile communication networks (2G, 3G, 4G, and 5G), Wireless local area networks, and/or WiFi (Wireless Fidelity) networks. In some embodiments, the
rf circuit504 may further include NFC (Near Field Communication) related circuits, which are not limited in this application.
The
display screen505 is used to display a UI (User Interface). The UI may include graphics, text, icons, video, and any combination thereof. When the
display screen505 is a touch display screen, the
display screen505 also has the ability to capture touch signals on or over the surface of the
display screen505. The touch signal may be input to the
processor501 as a control signal for processing. At this point, the
display screen505 may also be used to provide virtual buttons and/or a virtual keyboard, also referred to as soft buttons and/or a soft keyboard. In some embodiments, the
display screen505 may be one, providing the front panel of the terminal 500; in other embodiments, the display screens 505 may be at least two, respectively disposed on different surfaces of the terminal 500 or in a folded design; in still other embodiments, the
display505 may be a flexible display disposed on a curved surface or on a folded surface of the terminal 500. Even more, the
display screen505 can be arranged in a non-rectangular irregular figure, i.e. a shaped screen. The
Display screen505 may be made of LCD (Liquid Crystal Display), OLED (Organic Light-Emitting Diode), and other materials.
The
camera assembly506 is used to capture images or video. Optionally,
camera assembly506 includes a front camera and a rear camera. Generally, a front camera is disposed at a front panel of the terminal, and a rear camera is disposed at a rear surface of the terminal. In some embodiments, the number of the rear cameras is at least two, and each rear camera is any one of a main camera, a depth-of-field camera, a wide-angle camera and a telephoto camera, so that the main camera and the depth-of-field camera are fused to realize a background blurring function, and the main camera and the wide-angle camera are fused to realize panoramic shooting and VR (Virtual Reality) shooting functions or other fusion shooting functions. In some embodiments,
camera assembly506 may also include a flash. The flash lamp can be a monochrome temperature flash lamp or a bicolor temperature flash lamp. The double-color-temperature flash lamp is a combination of a warm-light flash lamp and a cold-light flash lamp, and can be used for light compensation at different color temperatures.
507 may include a microphone and a speaker. The microphone is used for collecting sound waves of a user and the environment, converting the sound waves into electric signals, and inputting the electric signals to the
processor501 for processing, or inputting the electric signals to the
radio frequency circuit504 to realize voice communication. For the purpose of stereo sound collection or noise reduction, a plurality of microphones may be provided at different portions of the terminal 500. The microphone may also be an array microphone or an omni-directional pick-up microphone. The speaker is used to convert electrical signals from the
processor501 or the
radio frequency circuit504 into sound waves. The loudspeaker can be a traditional film loudspeaker or a piezoelectric ceramic loudspeaker. When the speaker is a piezoelectric ceramic speaker, the speaker can be used for purposes such as converting an electric signal into a sound wave audible to a human being, or converting an electric signal into a sound wave inaudible to a human being to measure a distance. In some embodiments,
audio circuitry507 may also include a headphone jack.
The
positioning component508 is used for positioning the current geographic Location of the terminal 500 for navigation or LBS (Location Based Service). The
Positioning component508 may be a Positioning component based on the united states GPS (Global Positioning System), the chinese beidou System, the russian graves System, or the european union's galileo System.
509 is used to power the various components in
terminal500. The
power source509 may be alternating current, direct current, disposable or rechargeable. When
power supply509 includes a rechargeable battery, the rechargeable battery may support wired or wireless charging. The rechargeable battery may also be used to support fast charge technology.
In some embodiments, terminal 500 also includes one or more sensors 510. The one or more sensors 510 include, but are not limited to: acceleration sensor 511, gyro sensor 512, pressure sensor 513, fingerprint sensor 514, optical sensor 515, and proximity sensor 516.
The acceleration sensor 511 may detect the magnitude of acceleration on three coordinate axes of the coordinate system established with the terminal 500. For example, the acceleration sensor 511 may be used to detect components of the gravitational acceleration in three coordinate axes. The
processor501 may control the
touch screen505 to display the user interface in a landscape view or a portrait view according to the gravitational acceleration signal collected by the acceleration sensor 511. The acceleration sensor 511 may also be used for acquisition of motion data of a game or a user.
The gyro sensor 512 may detect a body direction and a rotation angle of the terminal 500, and the gyro sensor 512 may cooperate with the acceleration sensor 511 to acquire a 3D motion of the user on the
terminal500. The
processor501 may implement the following functions according to the data collected by the gyro sensor 512: motion sensing (such as changing the UI according to a user's tilting operation), image stabilization at the time of photographing, game control, and inertial navigation.
The pressure sensor 513 may be disposed on a side bezel of the terminal 500 and/or an underlying layer of the
touch display screen505. When the pressure sensor 513 is disposed on the side frame of the terminal 500, a user's holding signal of the terminal 500 may be detected, and the
processor501 performs left-right hand recognition or shortcut operation according to the holding signal collected by the pressure sensor 513. When the pressure sensor 513 is disposed at the lower layer of the
touch display screen505, the
processor501 controls the operability control on the UI interface according to the pressure operation of the user on the
touch display screen505. The operability control comprises at least one of a button control, a scroll bar control, an icon control and a menu control.
The fingerprint sensor 514 is used for collecting a fingerprint of the user, and the
processor501 identifies the identity of the user according to the fingerprint collected by the fingerprint sensor 514, or the fingerprint sensor 514 identifies the identity of the user according to the collected fingerprint. Upon recognizing that the user's identity is a trusted identity, the
processor501 authorizes the user to perform relevant sensitive operations including unlocking the screen, viewing encrypted information, downloading software, paying, and changing settings, etc. The fingerprint sensor 514 may be provided on the front, back, or side of the terminal 500. When a physical button or a vendor Logo is provided on the terminal 500, the fingerprint sensor 514 may be integrated with the physical button or the vendor Logo.
The optical sensor 515 is used to collect the ambient light intensity. In one embodiment, the
processor501 may control the display brightness of the
touch display screen505 based on the ambient light intensity collected by the optical sensor 515. Specifically, when the ambient light intensity is high, the display brightness of the
touch display screen505 is increased; when the ambient light intensity is low, the display brightness of the
touch display screen505 is turned down. In another embodiment,
processor501 may also dynamically adjust the shooting parameters of
camera head assembly506 based on the ambient light intensity collected by optical sensor 515.
A proximity sensor 516, also referred to as a distance sensor, is typically disposed on the front panel of the terminal 500. The proximity sensor 516 is used to collect the distance between the user and the front surface of the terminal 500. In one embodiment, when the proximity sensor 516 detects that the distance between the user and the front surface of the terminal 500 gradually decreases, the
processor501 controls the
touch display screen505 to switch from the bright screen state to the dark screen state; when the proximity sensor 516 detects that the distance between the user and the front surface of the terminal 500 becomes gradually larger, the
processor501 controls the
touch display screen505 to switch from the screen-rest state to the screen-on state.
Those skilled in the art will appreciate that the configuration shown in fig. 5 is not intended to be limiting of
terminal500 and may include more or fewer components than those shown, or some components may be combined, or a different arrangement of components may be used.
Fig. 6 is a schematic structural diagram of a server according to an embodiment of the present application, where the server may be a server according to the foregoing embodiment, and the
server600 may generate a relatively large difference due to different configurations or performances, and may include one or more processors (CPUs) 601 and one or
more memories602, where at least one instruction is stored in the
memories602, and the at least one instruction is loaded and executed by the
processors601 to implement the methods provided by the foregoing method embodiments. Of course, the server may also have components such as a wired or wireless network interface, a keyboard, and an input/output interface, so as to perform input/output, and the server may also include other components for implementing the functions of the device, which are not described herein again.
In an exemplary embodiment, a computer-readable storage medium, such as a memory including instructions executable by a processor in a terminal, is also provided to perform the method of training a search result scoring model in the above embodiments. For example, the computer-readable storage medium may be a Read-only Memory (ROM), a Random Access Memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like.
It will be understood by those skilled in the art that all or part of the steps for implementing the above embodiments may be implemented by hardware, or may be implemented by a program instructing relevant hardware, where the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a magnetic disk or an optical disk, etc.
The above description is only exemplary of the present application and should not be taken as limiting, as any modification, equivalent replacement, or improvement made within the spirit and principle of the present application should be included in the protection scope of the present application.
Claims (12)
1. A method of training a search result scoring model, the method comprising:
receiving a search request carrying first search information sent by a first terminal, and sending a plurality of search results corresponding to the first search information to the first terminal;
acquiring operation information of the first terminal on each search result after the plurality of search results are displayed, and determining an operation class corresponding to the operation information of each search result, wherein the operation information is used for indicating whether the first terminal triggers the search result after the search result is displayed for a first preset time or before a target function is triggered, and if the search result is triggered, the first terminal displays the time of a page corresponding to the search result and whether the target function is triggered through the page;
determining a reference score corresponding to each search result based on a pre-stored reference score corresponding to each operation class and an operation class corresponding to each search result;
and training a search result scoring model based on the plurality of search results and the reference score corresponding to each search result.
2. The method of claim 1, wherein the operation class comprises at least one of:
before reaching a first preset time or triggering a target function, the first terminal does not trigger the search result;
before reaching a first preset time or triggering a target function, the first terminal triggers the search result to display a corresponding page and closes the page, and the target function is not triggered through the page, wherein the display time of the page is less than a second preset time;
before reaching a first preset time or triggering a target function, the first terminal triggers the search result to display a corresponding page and close the page, and the target function is not triggered by the page, wherein the display time of the page is longer than a second preset time and shorter than a third preset time;
before the first preset duration is reached, the first terminal triggers the search result to display a corresponding page, and triggers the target function through the page;
the first preset time length is longer than a third preset time length, and the third preset time length is longer than the second preset time length.
3. The method of claim 1, wherein training a search result scoring model based on the plurality of search results and the respective base scores for each search result comprises:
pairing the plurality of search results to obtain at least one search result pair, wherein two search results in each search result pair correspond to different operation classes;
respectively inputting each search result into a search result scoring model to obtain an actual score corresponding to each search result, and determining the actual difference of each search result pair based on the actual scores corresponding to two search results in each search result pair;
determining the benchmark difference degree of each search result pair based on the benchmark scores corresponding to the two search results in each search result pair;
training the search result scoring model based on the actual difference and the reference difference of each search result pair.
4. The method of claim 3, wherein determining the actual degree of difference for each search result pair based on the actual scores for the two search results in each search result pair comprises:
and determining the actual difference degree of each search result pair based on the magnitude relation of the actual scores corresponding to the two search results in each search result pair.
5. The method of claim 3, wherein determining the baseline degree of difference for each search result pair based on the baseline scores for the two search results in each search result pair comprises:
and determining the benchmark difference degree of each search result pair based on the magnitude relation of the benchmark scores corresponding to the two search results of each search result pair.
6. The method of claim 5, wherein the two search results in each pair of search results are arranged in a specified order, and wherein determining the base difference degree for each pair of search results based on the magnitude relationship of the base scores corresponding to the two search results for each pair of search results comprises:
for each search result pair, if the reference score corresponding to the search result ranked in the front of the search result pair is greater than the reference score corresponding to the search result ranked in the back of the search result pair, determining that the reference difference degree corresponding to the search result pair is 1, and if the reference score corresponding to the search result ranked in the front of the search result pair is less than the reference score corresponding to the search result ranked in the back of the search result pair, determining that the reference difference degree corresponding to the search result pair is 0.
7. The method of claim 4, wherein determining the actual degree of difference for each search result pair based on the magnitude relationship of the actual scores corresponding to two search results in each search result pair comprises:
for each search result pair, by formula
Determining the actual degree of dissimilarity g (item) for each search result pairi),f(itemj) Wherein, the f (item)i) The actual score of the top ranked search result of the search result pair, f (item)j) The actual score of the top ranked search result of the search result pair is assigned.
8. The method of claim 3, wherein training the search result scoring model based on the actual and baseline degrees of difference for each search result pair comprises:
and for each search result pair, inputting the actual difference and the reference difference corresponding to the search result pair into a loss function to obtain an adjustment value of a parameter in the search result scoring model, and updating the search result scoring model based on the adjustment value of the parameter.
9. The method of claim 1, wherein after training the search result scoring model based on the plurality of search results and the respective base scores for each of the plurality of search results, further comprising:
receiving a search request carrying second search information sent by a second terminal;
obtaining a plurality of search results corresponding to the second search information;
determining a score corresponding to each search result based on the second search information, each search result and the trained search result scoring model;
ranking the search results based on the score corresponding to each search result to obtain a search result list;
and sending the search result list to the second terminal.
10. An apparatus for training a search result scoring model, the apparatus comprising:
the device comprises a sending module, a searching module and a searching module, wherein the sending module is used for receiving a searching request which is sent by a first terminal and carries first searching information and sending a plurality of searching results corresponding to the first searching information to the first terminal;
an obtaining module, configured to obtain operation information of the first terminal for each search result after the plurality of search results are displayed, and determine an operation class corresponding to the operation information of each search result, where the operation information is used to indicate whether the first terminal triggers the search result after the search result is displayed for a first preset time or before a target function is triggered, and if the search result is triggered, the first terminal triggers a display time of a page corresponding to the search result and whether the target function is triggered by the page;
the determining module is used for determining the reference score corresponding to each search result based on the pre-stored reference score corresponding to each operation class and the operation class corresponding to each search result;
and the training module is used for training the search result scoring model based on the plurality of search results and the reference score corresponding to each search result.
11. A computer device comprising a processor and a memory, the memory having stored therein at least one instruction that is loaded and executed by the processor to perform operations performed by the method of training a search result scoring model according to any one of claims 1 to 9.
12. A computer-readable storage medium having stored therein at least one instruction which is loaded and executed by a processor to perform operations performed by the method of training a search result scoring model according to any one of claims 1 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010941415.4A CN112100528B (en) | 2020-09-09 | 2020-09-09 | Method, device, equipment and medium for training search result scoring model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010941415.4A CN112100528B (en) | 2020-09-09 | 2020-09-09 | Method, device, equipment and medium for training search result scoring model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112100528A true CN112100528A (en) | 2020-12-18 |
| CN112100528B CN112100528B (en) | 2024-06-28 |
Family
ID=73752491
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010941415.4A Active CN112100528B (en) | 2020-09-09 | 2020-09-09 | Method, device, equipment and medium for training search result scoring model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112100528B (en) |
Cited By (2)
* Cited by examiner, † Cited by third party| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112948690A (en) * | 2021-03-29 | 2021-06-11 | 北京三快在线科技有限公司 | Search method, device, equipment and storage medium |
| CN114817521A (en) * | 2021-01-20 | 2022-07-29 | 华为技术有限公司 | Searching method and electronic equipment |
Citations (7)
* Cited by examiner, † Cited by third party| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017071578A1 (en) * | 2015-10-27 | 2017-05-04 | 广州神马移动信息科技有限公司 | Searching method, apparatus and system, and method for determining search result order adjustment basis |
| CN106777248A (en) * | 2016-12-27 | 2017-05-31 | 努比亚技术有限公司 | A kind of search engine test evaluation method and apparatus |
| CN107506402A (en) * | 2017-08-03 | 2017-12-22 | 北京百度网讯科技有限公司 | Sort method, device, equipment and the computer-readable recording medium of search result |
| CN108345601A (en) * | 2017-01-23 | 2018-07-31 | 腾讯科技（深圳）有限公司 | Search result ordering method and device |
| CN109086417A (en) * | 2018-08-06 | 2018-12-25 | 清华大学 | Search evaluation method and device |
| US20190018900A1 (en) * | 2017-07-14 | 2019-01-17 | Alibaba Group Holding Limited | Method and Apparatus for Displaying Search Results |
| CN110489638A (en) * | 2019-07-08 | 2019-11-22 | 广州视源电子科技股份有限公司 | A kind of searching method, device, server, system and storage medium |
-
2020
- 2020-09-09 CN CN202010941415.4A patent/CN112100528B/en active Active
Patent Citations (7)
* Cited by examiner, † Cited by third party| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017071578A1 (en) * | 2015-10-27 | 2017-05-04 | 广州神马移动信息科技有限公司 | Searching method, apparatus and system, and method for determining search result order adjustment basis |
| CN106777248A (en) * | 2016-12-27 | 2017-05-31 | 努比亚技术有限公司 | A kind of search engine test evaluation method and apparatus |
| CN108345601A (en) * | 2017-01-23 | 2018-07-31 | 腾讯科技（深圳）有限公司 | Search result ordering method and device |
| US20190018900A1 (en) * | 2017-07-14 | 2019-01-17 | Alibaba Group Holding Limited | Method and Apparatus for Displaying Search Results |
| CN107506402A (en) * | 2017-08-03 | 2017-12-22 | 北京百度网讯科技有限公司 | Sort method, device, equipment and the computer-readable recording medium of search result |
| CN109086417A (en) * | 2018-08-06 | 2018-12-25 | 清华大学 | Search evaluation method and device |
| CN110489638A (en) * | 2019-07-08 | 2019-11-22 | 广州视源电子科技股份有限公司 | A kind of searching method, device, server, system and storage medium |
Non-Patent Citations (1)
* Cited by examiner, † Cited by third party| Title |
|---|
| 金祖旭 等: "基于用户反馈的搜索引擎排名算法", 计算机系统应用, no. 11, 15 November 2010 (2010-11-15) * |
Cited By (3)
* Cited by examiner, † Cited by third party| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114817521A (en) * | 2021-01-20 | 2022-07-29 | 华为技术有限公司 | Searching method and electronic equipment |
| CN114817521B (en) * | 2021-01-20 | 2024-05-14 | 华为技术有限公司 | Searching method and electronic equipment |
| CN112948690A (en) * | 2021-03-29 | 2021-06-11 | 北京三快在线科技有限公司 | Search method, device, equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112100528B (en) | 2024-06-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112162671B (en) | 2022-03-25 | Live broadcast data processing method and device, electronic equipment and storage medium |
| CN110674022B (en) | 2021-04-27 | Behavior data acquisition method and device and storage medium |
| CN110868626B (en) | 2021-06-11 | Method and device for preloading content data |
| CN110471858B (en) | 2023-09-01 | Application program testing method, device and storage medium |
| CN110865754B (en) | 2020-09-22 | Information display method and device and terminal |
| CN111079012A (en) | 2020-04-28 | Live broadcast room recommendation method and device, storage medium and terminal |
| CN108737897B (en) | 2020-11-10 | Video playing method, device, equipment and storage medium |
| CN109327608B (en) | 2021-01-22 | Song sharing method, terminal, server and system |
| CN109451343A (en) | 2019-03-08 | Video sharing method, apparatus, terminal and storage medium |
| CN111782950B (en) | 2024-07-23 | Sample data set acquisition method, device, equipment and storage medium |
| CN110248236B (en) | 2022-02-18 | Video playing method, device, terminal and storage medium |
| CN109922356B (en) | 2021-07-09 | Video recommendation method and device and computer-readable storage medium |
| WO2022057435A1 (en) | 2022-03-24 | Search-based question answering method, and storage medium |
| CN110147503B (en) | 2021-08-17 | Information issuing method and device, computer equipment and storage medium |
| CN110413837A (en) | 2019-11-05 | Video recommendation method and device |
| CN112667835A (en) | 2021-04-16 | Work processing method and device, electronic equipment and storage medium |
| CN111796990A (en) | 2020-10-20 | Resource display method, device, terminal and storage medium |
| CN112100528B (en) | 2024-06-28 | Method, device, equipment and medium for training search result scoring model |
| CN111192072A (en) | 2020-05-22 | User grouping method and device and storage medium |
| CN109388732B (en) | 2022-02-25 | Music map generating and displaying method, device and storage medium |
| CN111563201A (en) | 2020-08-21 | Content pushing method, device, server and storage medium |
| CN111523876A (en) | 2020-08-11 | Payment mode display method, device and system and storage medium |
| CN110992954A (en) | 2020-04-10 | Method, device, equipment and storage medium for voice recognition |
| CN110399193A (en) | 2019-11-01 | Guide data sending method, device, terminal, server and storage medium |
| CN110851435B (en) | 2022-02-11 | Data storage method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 2020-12-18 | PB01 | Publication | |
| 2020-12-18 | PB01 | Publication | |
| 2021-01-15 | SE01 | Entry into force of request for substantive examination | |
| 2021-01-15 | SE01 | Entry into force of request for substantive examination | |
| 2024-06-28 | GR01 | Patent grant | |
| 2024-06-28 | GR01 | Patent grant |