CN113420056B - Behavior data processing method and device, electronic equipment and storage medium - Google Patents

Behavior data processing method and device, electronic equipment and storage medium Download PDF

Info

Publication number
CN113420056B
CN113420056B CN202110529224.1A CN202110529224A CN113420056B CN 113420056 B CN113420056 B CN 113420056B CN 202110529224 A CN202110529224 A CN 202110529224A CN 113420056 B CN113420056 B CN 113420056B
Authority
CN
China
Prior art keywords
information
search result
search
similarity
result
Prior art date
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.)
Active
Application number
CN202110529224.1A
Other languages
Chinese (zh)
Other versions
CN113420056A (en
Inventor
张志伟
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Dajia Internet Information Technology Co Ltd
Original Assignee
Beijing Dajia Internet Information Technology Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Beijing Dajia Internet Information Technology Co Ltd filed Critical Beijing Dajia Internet Information Technology Co Ltd
Priority to CN202110529224.1A priority Critical patent/CN113420056B/en
Publication of CN113420056A publication Critical patent/CN113420056A/en
Application granted granted Critical
Publication of CN113420056B publication Critical patent/CN113420056B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24—Querying
    • G06F16/245—Query processing
    • G06F16/2457—Query processing with adaptation to user needs
    • 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
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Databases & Information Systems (AREA)
  • Computational Linguistics (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The disclosure relates to a behavior data processing method, a device, an electronic device and a storage medium, wherein the behavior data processing method comprises the following steps: acquiring at least one search result corresponding to the target search information; determining a similarity between the target search information and the at least one search result; determining object behavior data corresponding to the at least one search result; and correcting the object behavior data corresponding to the at least one search result based on the similarity between the target search information and the at least one search result to obtain the target object behavior data corresponding to the at least one search result. By utilizing the technical scheme provided by the embodiment of the invention, the deviation existing when the preference degree of the user is described through the target object behavior data can be reduced.

Description

Behavior data processing method and device, electronic equipment and storage medium
Technical Field
The disclosure relates to the technical field of internet, and in particular relates to a behavior data processing method, a behavior data processing device, electronic equipment and a storage medium.
Background
In the related art, in a search scenario, it is often required to analyze the behavior of a user using posterior data, for example, by searching some videos with higher attractiveness under a certain search term through the posterior data of click through rate (ClickThroughRate, CTR), so as to characterize the preference degree of the user according to some videos with higher attractiveness. The posterior data refers to data generated when a user uses an Application (APP).
But for a certain search term, any search system cannot guarantee that the returned videos are very relevant, and this phenomenon is particularly serious in long-tail search terms. In the case that the returned result is an irrelevant video, the posterior data cannot accurately reflect the preference of the user, that is, the calculation method of the posterior data (for example, CTR) in the related art has a certain deviation when describing the preference degree of the user.
Disclosure of Invention
The disclosure provides a behavior data processing method, a behavior data processing device, electronic equipment and a storage medium, so as to at least solve the problem that in the related art, under the condition that returned results are irrelevant, a certain deviation exists in a calculation method of posterior data when describing the preference degree of a user. The technical scheme of the present disclosure is as follows:
according to a first aspect of an embodiment of the present disclosure, there is provided a behavior data processing method, including:
acquiring at least one search result corresponding to the target search information;
determining a similarity between the target search information and the at least one search result;
determining object behavior data corresponding to the at least one search result;
and correcting the object behavior data corresponding to the at least one search result based on the similarity between the target search information and the at least one search result to obtain the target object behavior data corresponding to the at least one search result.
In an exemplary embodiment, the correcting the object behavior data corresponding to the at least one search result based on the similarity between the target search information and the at least one search result to obtain the target object behavior data corresponding to the at least one search result includes:
and taking the similarity between the target search information and the at least one search result as a weight factor, and carrying out weighting processing on the object behavior data corresponding to the at least one search result based on the weight factor to obtain the target object behavior data corresponding to the at least one search result.
In an exemplary embodiment, if the target object behavior data corresponding to the at least one search result is a preset number, after the correcting the object behavior data corresponding to the at least one search result based on the similarity between the target search information and the at least one search result to obtain the target object behavior data corresponding to the at least one search result, the behavior data processing method further includes:
determining a ranking parameter corresponding to the at least one search result based on a preset number of target object behavior data corresponding to the at least one search result;
And ordering the at least one search result based on the ordering parameter corresponding to the at least one search result to obtain an ordering sequence corresponding to the at least one search result.
In an exemplary embodiment, the ranking the at least one search result based on the ranking parameter corresponding to the at least one search result to obtain the ranking sequence corresponding to the at least one search result includes:
filtering search results with similarity smaller than or equal to a similarity threshold value from the at least one search result, and obtaining filtered search results;
and ordering the filtered search results based on ordering parameters corresponding to the filtered search results to obtain the ordering sequence.
In an exemplary embodiment, the determining the similarity between the target search information and the at least one search result includes:
extracting features of the target search information to obtain a first information feature extraction result corresponding to the target search information;
performing feature extraction on the at least one search result to obtain a second information feature extraction result corresponding to the at least one search result;
And performing similarity matching processing on the first information feature extraction result and the second information feature extraction result to obtain the similarity between the target search information and the at least one search result, wherein the first information feature extraction result and the second information feature extraction result are in the same dimension.
In an exemplary embodiment, the performing similarity matching on the first information feature extraction result and the second information feature extraction result to obtain the similarity between the target search information and the at least one search result includes:
performing similarity matching processing on the first information feature extraction result and the second information feature extraction result based on a similarity model to obtain similarity between the target search information and the at least one search result;
the generation mode of the similarity model comprises the following steps:
obtaining a sample data set, wherein the sample data set comprises at least one sample data, the at least one sample data has a similarity label, the at least one sample data comprises sample search information and a sample search result corresponding to the sample search information, and the similarity label represents similarity between the sample search information and the sample search result;
Extracting features of the sample searching information to obtain a first sample information feature extraction result corresponding to the sample searching information;
extracting features of the sample search results to obtain second sample information feature extraction results corresponding to the sample search results;
performing similarity training on a neural network based on the first sample information feature extraction result and the second sample information feature extraction result to obtain a similarity prediction result between the first sample information feature extraction result and the second sample information feature extraction result;
determining loss data based on the similarity prediction result and the similarity label;
and training the neural network based on the loss data to obtain the similarity model.
In an exemplary embodiment, the determining the object behavior data corresponding to the at least one search result includes:
acquiring log data corresponding to the at least one search result;
and counting the log data corresponding to the at least one search result to obtain object behavior data corresponding to the at least one search result.
In an exemplary embodiment, the counting the log data corresponding to the at least one search result to obtain the object behavior data corresponding to the at least one search result includes:
Acquiring display information and corresponding operation information of at least one search result in preset time from log data corresponding to the at least one search result;
and determining object behavior data corresponding to the at least one search result based on the display information and the corresponding operation information of the at least one search result in the preset time.
According to a second aspect of embodiments of the present disclosure, there is provided a behavioural data processing device, comprising;
the search result acquisition module is configured to execute at least one search result corresponding to the acquisition target search information;
a similarity determination module configured to perform a determination of a similarity between the target search information and the at least one search result;
an object behavior data determining module configured to perform determining object behavior data corresponding to the at least one search result;
and the correction module is configured to execute correction on the object behavior data corresponding to the at least one search result based on the similarity between the target search information and the at least one search result, so as to obtain the target object behavior data corresponding to the at least one search result.
In an exemplary embodiment, the correction module is configured to perform weighting processing on object behavior data corresponding to the at least one search result based on a weight factor by using similarity between the target search information and the at least one search result as the weight factor, so as to obtain the target object behavior data corresponding to the at least one search result.
In an exemplary embodiment, the target object behavior data corresponding to the at least one search result is a preset number, and the behavior data processing device further includes:
a ranking parameter determining module configured to perform determining a ranking parameter corresponding to the at least one search result based on a preset number of target object behavior data corresponding to the at least one search result;
and the ordering sequence determining module is configured to execute ordering of the at least one search result based on the ordering parameter corresponding to the at least one search result, so as to obtain an ordering sequence corresponding to the at least one search result.
In an exemplary embodiment, the ordered sequence determination module includes:
a filtering unit configured to perform filtering of search results with a similarity with the target search information being less than or equal to a similarity threshold value from the at least one search result, to obtain filtered search results;
And the sorting unit is configured to sort the filtered search results based on the sorting parameters corresponding to the filtered search results to obtain the sorting sequence.
In an exemplary embodiment, the similarity determination module includes:
the first feature extraction unit is configured to perform feature extraction on the target search information to obtain a first information feature extraction result corresponding to the target search information;
a second feature extraction unit configured to perform feature extraction on the at least one search result to obtain a second information feature extraction result corresponding to the at least one search result;
and the matching processing unit is configured to perform similarity matching processing on the first information feature extraction result and the second information feature extraction result to obtain similarity between the target search information and the at least one search result, wherein the first information feature extraction result and the second information feature extraction result are in the same dimension.
In an exemplary embodiment, the matching processing unit is configured to perform similarity matching processing on the first information feature extraction result and the second information feature extraction result based on a similarity model, so as to obtain similarity between the target search information and the at least one search result;
The behavior data processing apparatus further includes:
a sample data set acquisition module configured to perform acquiring a sample data set, the sample data set including at least one sample data, the at least one sample data having a similarity tag, the at least one sample data including sample search information and sample search results corresponding to the sample search information, the similarity tag characterizing a similarity between the sample search information and the sample search results;
the first sample characteristic extraction module is configured to perform characteristic extraction on the sample search information to obtain a first sample information characteristic extraction result corresponding to the sample search information;
the second sample feature extraction module is configured to perform feature extraction on the sample search result to obtain a second sample information feature extraction result corresponding to the sample search result;
the training module is configured to perform similarity training on the neural network based on the first sample information feature extraction result and the second sample information feature extraction result, and obtain a similarity prediction result between the first sample information feature extraction result and the second sample information feature extraction result;
A loss data determination module configured to perform determining loss data based on the similarity prediction result and the similarity label;
a similarity model determination module configured to perform training of the neural network based on the loss data to obtain the similarity model.
In an exemplary embodiment, the object behavior data determination module includes:
a log data obtaining unit configured to obtain log data corresponding to the at least one search result;
and the statistics unit is configured to perform statistics on the log data corresponding to the at least one search result to obtain object behavior data corresponding to the at least one search result.
In an exemplary embodiment, the statistics unit includes:
an information obtaining subunit, configured to obtain display information and corresponding operation information of the at least one search result in a preset time from log data corresponding to the at least one search result;
and the object behavior data determining subunit is configured to determine object behavior data corresponding to the at least one search result based on the display information and the corresponding operation information of the at least one search result in the preset time.
According to a third aspect of embodiments of the present disclosure, there is provided an electronic device comprising;
a processor;
a memory for storing the processor-executable instructions;
wherein the processor is configured to execute the instructions to implement the behavioural data processing method as described in any one of the embodiments above.
According to a fourth aspect of embodiments of the present disclosure, there is provided a computer readable storage medium, which when executed by a processor of an electronic device, causes the electronic device to perform a behavioural data processing method as described in any one of the preceding embodiments.
According to a fifth aspect of the disclosed embodiments, there is provided a computer program product comprising a computer program which, when executed by a processor, implements the behavioural data processing method according to any one of the embodiments described above.
The technical scheme provided by the embodiment of the disclosure at least brings the following beneficial effects:
the method comprises the steps of obtaining at least one search result corresponding to target search information, determining the similarity between the target search information and the at least one search result, and correcting object behavior data corresponding to the at least one search result according to the similarity between the target search information and the at least one search result to obtain target object behavior data corresponding to the at least one search result. Because the corresponding object behavior data can be corrected through the similarity between the target search information and at least one search result (namely, the similarity determined through priori knowledge is corrected for the posterior object data behavior), the corrected target object behavior data can accurately reflect the preference of the user, and further, the deviation existing when the preference degree of the user is represented through the target object behavior data is reduced.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the disclosure and together with the description, serve to explain the principles of the disclosure and do not constitute an undue limitation on the disclosure.
FIG. 1 is an application environment diagram illustrating a behavior data processing method according to an example embodiment.
FIG. 2 is a flowchart illustrating a behavioral data processing method according to an example embodiment.
FIG. 3 is a flowchart illustrating another behavioral data process according to an example embodiment.
FIG. 4 is a flowchart illustrating a manner in which a similarity model is generated, according to an example embodiment.
FIG. 5 is a flowchart illustrating another behavioral data processing method according to an example embodiment.
FIG. 6 is a flowchart illustrating another behavioral data processing method according to an example embodiment.
FIG. 7 is a block diagram of a behavioural data processing apparatus, according to an example embodiment.
FIG. 8 is a block diagram of an electronic device for behavioral data processing according to an example embodiment.
Detailed Description
In order to enable those skilled in the art to better understand the technical solutions of the present disclosure, the technical solutions of the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings.
It should be noted that the terms "first," "second," and the like in the description and claims of the present disclosure and in the foregoing figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the disclosure described herein may be capable of operation in sequences other than those illustrated or described herein. The implementations described in the following exemplary examples are not representative of all implementations consistent with the present disclosure. Rather, they are merely examples of apparatus and methods consistent with some aspects of the present disclosure as detailed in the accompanying claims.
Referring to fig. 1, fig. 1 is a diagram illustrating an application environment of a behavior data processing method according to an exemplary embodiment, where the application environment may include a client 01 and a server 02.
Wherein the client 01 may be configured to collect target search information and send the target search information to the server 02. Optionally, the client 01 may include a terminal device such as a smart phone, a desktop computer, a tablet computer, a notebook computer, a digital assistant, an augmented reality (augmented reality, AR)/Virtual Reality (VR) device, a smart wearable device, or may include a server that operates independently, or a distributed server, or a server cluster that is composed of multiple servers.
The server 02 may be used to provide background services for the client 01. For example, the server 02 may process the target search information sent by the client 01, obtain at least one search result corresponding to the target search information, and sequence the at least one search result and return the at least one search result to the client 01. Optionally, the server 02 may be a server cluster or a distributed system including an independent physical server or a plurality of physical servers, and may also be a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (Content Delivery Network, CDN), and basic cloud computing services such as big data and artificial intelligence platforms.
Fig. 2 is a flowchart of a behavior data processing method according to an exemplary embodiment, as shown in fig. 2, with the behavior data processing method being used in the server 02 shown in fig. 1 for explanation, including the steps of:
in step S11, at least one search result corresponding to the target search information is acquired.
In the embodiment of the disclosure, when a user corresponding to the client 01 searches in a search system of the APP, the search system of the APP may be opened, and target search information is input in an information input interface of the search system, so as to trigger an information search instruction, and the server 02 searches at least one search result corresponding to the target search information from the database in response to the search instruction.
Illustratively, the APP may be, but is not limited to, a short video APP, a news APP, a game APP, and the like.
The target search information (query) may be text, pictures, voice, etc., by way of example, but is not limited thereto.
Illustratively, the search results may be, but are not limited to, video, pictures, text, and the like.
Specifically, taking the search result as a picture (photo) as an example, in the above step S11, the data form of the obtained at least one search result may be < query, photo >.
In step S13, a similarity between the target search information and the at least one search result is determined. Illustratively, the similarity refers to a similarity between the target search information and the at least one search result, that is, a degree of matching between the target search information and the at least one search result.
Specifically, the similarity may be cosine similarity. The cosine similarity is calculated by calculating an included angle cosine value between a vector corresponding to the target search information and a vector corresponding to the at least one search result, so as to evaluate the similarity between the vector corresponding to the target search information and the at least one search result.
In some embodiments, the similarity between the two may also be measured by other means, such as by pearson correlation coefficients. Wherein, the pearson correlation coefficient is used for representing the similarity by using the linear correlation of the vector.
In a particular embodiment, FIG. 3 is a flow chart illustrating another behavioral data process according to an example embodiment. As shown in fig. 3, in the step S13, the determining the similarity between the target search information and the at least one search result may include:
in step S131, feature extraction is performed on the target search information, so as to obtain a first information feature extraction result corresponding to the target search information.
In step S133, feature extraction is performed on the at least one search result, so as to obtain a second information feature extraction result corresponding to the at least one search result.
In step S135, a similarity matching process is performed on the first information feature extraction result and the second information feature extraction result, so as to obtain a similarity between the target search information and the one search result, where the first information feature extraction result and the second information feature extraction result are in the same dimension.
Illustratively, when the target search information is text, a text feature extraction model may be used in the above step S131 text And extracting text features of the target search information to obtain the first information feature extraction result. Alternatively, the model text Any open-source chinese word Vector model may be used, such as chinese word2Vector.
Illustratively, in the case where the search result is a picture, in the above step S133, a convolutional neural network model of the extracted image may be used image And extracting image features of the search result to obtain the second information feature extraction result. Alternatively, the model image May be an open source model, such as an ImageNet model, a proprietary model, etc.
In the embodiment of the disclosure, instead of performing similarity matching directly according to the target search information and at least one search result, feature extraction is performed on the target search information and the at least one search result through the existing feature extraction model, and then similarity matching processing is performed on the basis of the extracted information feature extraction result, so that not only can the utilization rate of the existing model be improved, the cost for determining the similarity be reduced, but also the accuracy for determining the similarity between the target search information and the at least one search result can be improved.
In an alternative embodiment, the step S135 may include:
and performing similarity matching processing on the first information feature extraction result and the second information feature extraction result based on the similarity model to obtain similarity between the target search information and the at least one search result.
In an alternative embodiment, as shown in fig. 4, fig. 4 is a flow chart illustrating a manner of generating a similarity model according to an exemplary embodiment. Accordingly, the generation mode of the similarity model may include the following steps:
in step S21, a sample data set is obtained, where the sample data set includes at least one sample data, the at least one sample data has a similarity tag, the at least one sample data includes sample search information and a sample search result corresponding to the sample search information, and the similarity tag characterizes a similarity between the sample search information and the sample search result.
In step S23, feature extraction is performed on the sample search information, so as to obtain a first sample information feature extraction result corresponding to the sample search information.
In step S25, feature extraction is performed on the sample search result, so as to obtain a second sample information feature extraction result corresponding to the sample search result.
In step S27, similarity training is performed on the neural network based on the first sample information feature extraction result and the second sample information feature extraction result, so as to obtain a similarity prediction result between the first sample information feature extraction result and the second sample information feature extraction result.
In step S29, loss data is determined based on the similarity prediction result and the similarity label.
In step S211, the neural network is trained based on the loss data to obtain the similarity model.
In the embodiment of the disclosure, the neural network may be trained through the sample data set labeled with the similarity label, so as to obtain the similarity model.
For example, with the search result being a picture, for any sample data i in the above step S21, a specific data format may be:
<query i ,photo i ,label i >，
wherein query is i To search information for samples i For sample search results, label i Is a similarity label.
Illustratively, when the sample search information is text, a text feature extraction model may be used in step S23 described above text And extracting text features of the sample search information to obtain the first sample information feature extraction result. Alternatively, the model text Any open-source chinese word Vector model may be used, such as chinese word2Vector.
Illustratively, in the case where the sample search result is a picture, in the above step S25, a convolutional neural network model of the extracted image may be used image And extracting image features from the sample search result to obtain the second sample information feature extraction result. Alternatively, the model image May be an open source model, such as an ImageNet model, a proprietary model, etc. Wherein ImageNet is a large visual database for visual object recognition software research.
Illustratively, the neural network in step S27 may be a recurrent neural network (Recurrent Neural Network, RNN), a convolutional neural network (Convolutional Neural Networks, CNN), or the like. And reducing the dimension of the first sample information characteristic extraction result and the second sample information characteristic extraction result to the same dimension through a neural network. And performing similarity training on the neural network through the first sample information feature extraction result and the second sample information feature extraction result after the dimension reduction to obtain a similarity prediction result.
The similarity prediction result may be a cosine similarity prediction result, for example. The similarity can of course also be measured in other ways, for example by pearson correlation coefficients.
Illustratively, in step S29-step S211, the similarity model may be trained using cross entropy loss according to the similarity prediction result and the similarity label.
In the embodiment of the disclosure, before training the neural network, the prior sample search information and the sample search result are subjected to feature extraction through the existing feature extraction model, and the neural network is trained based on the extracted feature extraction result, so that the utilization rate of the existing model can be improved, the training cost of the similarity model can be reduced, and the prior similarity model obtained through training can be ensured to have higher similarity matching precision.
In step S15, object behavior data corresponding to the at least one search result is determined.
In an optional embodiment, the determining the object behavior data corresponding to the at least one search result may include the following steps:
and acquiring log data corresponding to the at least one search result.
And counting the log data corresponding to the at least one search result to obtain object behavior data corresponding to the at least one search result.
In an alternative embodiment, the log data corresponding to the at least one search result may be posterior data generated when the user corresponding to the client performs an operation (such as clicking, praying, focusing, etc.) on the at least one search result.
In a specific embodiment, the posterior data may include log data recorded by the log itself, and statistics may be performed on log data (such as clicking, praise, attention, etc. data) corresponding to at least one search result, so as to obtain object behavior data corresponding to the at least one search result. Because the log data recorded by the log itself can truly reflect the operation condition of the user, the object behavior data corresponding to at least one search result is counted based on the log data, and the accuracy of determining the object behavior data corresponding to at least one search result can be ensured.
In another specific embodiment, the posterior data may include log data recorded by the log itself, or object behavior data counted according to the log data. Accordingly, the specific form of the posterior data may be as follows:
<query,userid,photoid,click,like,follow,..,date>，
wherein query refers to target search information, userid refers to user identity, photo refers to identification of search results, click refers to click, like refers to praise, follow refers to attention, and date refers to time. In this case, the statistically obtained object behavior data (including click rate, praise rate, attention rate, etc.) may be directly obtained from the posterior data.
In a specific embodiment, the counting the log data corresponding to the at least one search result to obtain the object behavior data corresponding to the at least one search result may include:
and acquiring display information and corresponding operation information of the at least one search result in a preset time from log data corresponding to the at least one search result.
And determining object behavior data corresponding to the at least one search result based on the display information and the corresponding operation information of the at least one search result in the preset time.
Illustratively, the "presentation information" refers to the number of presentations of the at least one search result presented by the client in a preset time. Accordingly, the "operation information" refers to operation information (such as click, praise, attention, etc. information) that the at least one search result is operated by the user corresponding to the client in a preset time.
For example, the object behavior data corresponding to the at least one search result may be determined according to a ratio of the operation information and the corresponding presentation information of the at least one search result within a preset time.
In a specific embodiment, taking at least one search result as a picture, and taking CTR as object behavior data corresponding to the at least one search result as an example, the above description is given of determining the object behavior data corresponding to the at least one search result:
assuming that at least one search result is a plurality of pictures, and the data form of each picture is < query-photo >, the calculation formula of CTR of each picture may be as follows:
where #SHOW indicates the number of times the query-photo pair is displayed at a preset time date, i refers to one sample (i.e., one query-photo pair), user refers to the user clicking on the query-photo, and click refers to the i sample being clicked.
In another specific embodiment, taking at least one search result as a picture, and taking an example that object behavior data corresponding to the at least one search result is a praise rate (ltr), the above-mentioned determining the object behavior data corresponding to the at least one search result is described as follows:
assuming that at least one search result is a plurality of pictures, and the data form of each picture is < query-photo >, the calculation formula of ltr of each picture can be as follows:
where #SHOW indicates the number of times the query-photo pair is displayed at a predetermined time date, i refers to one sample (i.e., one query-photo pair), user refers to the user clicking on the query-photo, like refers to the i sample being liked.
In another specific embodiment, taking at least one search result as a picture, and taking object behavior data corresponding to the at least one search result as a attention rate (ftr) as an example, the above description is given of determining the object behavior data corresponding to the at least one search result:
assuming that at least one search result is a plurality of pictures, and the data form of each picture is < query-photo >, the calculation formula of ftr of each picture may be as follows:
where #SHOW indicates the number of times the query-photo pair is displayed at a predetermined time date, i refers to one sample (i.e., one query-photo), user refers to the user clicking on the query-photo, and follow refers to the i sample being of interest.
Assuming that at least one search result is 10 pictures, statistics can be performed on clicking, praise, attention and other data corresponding to the 10 pictures according to log data corresponding to each of the 10 pictures, so as to obtain object behavior data (including clicking rate data, praise rate data, attention rate data and the like) corresponding to each of the 10 pictures.
According to the embodiment of the disclosure, according to the display information and the corresponding operation information of at least one search result recorded in the log data within the preset time, the accuracy of determining the object behavior data corresponding to the at least one search result can be further improved.
In step S17, the object behavior data corresponding to the at least one search result is modified based on the similarity between the target search information and the at least one search result, so as to obtain the target object behavior data corresponding to the at least one search result.
In an alternative embodiment, continuing to fig. 3, in step S17, correcting the object behavior data corresponding to the at least one search result based on the similarity between the target search information and the at least one search result to obtain the target object behavior data corresponding to the at least one search result may include: and weighting the object behavior data corresponding to the at least one search result based on the weight factor by taking the similarity between the target search information and the at least one search result as the weight factor to obtain the target object behavior data corresponding to the at least one search result.
Wherein the weighting process refers to "multiplying by a weight". Alternatively, the similarity between the target search information and the at least one search result may be used as a weight factor. Accordingly, in a specific embodiment, the correcting the object behavior data corresponding to the at least one search result based on the similarity between the target search information and the at least one search result to obtain the target object behavior data corresponding to the at least one search result may include: and calculating the similarity between the target search information and the at least one search result, and obtaining the product of the object behavior data corresponding to the at least one search result.
In a specific embodiment, taking at least one search result as a picture, and taking CTR as object behavior data corresponding to the at least one search result as an example, the similarity between the target search information and the at least one search result is taken as a weight factor, and weighting the object behavior data corresponding to the at least one search result based on the weight factor to obtain the target object behavior data corresponding to the at least one search result for explanation:
assuming that at least one search result is a plurality of pictures, and the data form of each picture is < query-photo >, the calculation formula for correcting the CTR of each picture through the corresponding similarity may be as follows:
wherein sim is qp Refers to similarity, sim qp Click refers to "sim qp * click ", i.e." sim qp Product of click).
In another specific embodiment, taking at least one search result as a picture and object behavior data corresponding to the at least one search result as ltr as an example, weighting the object behavior data corresponding to the at least one search result based on the weighting factor by taking similarity between the target search information and the at least one search result as the weighting factor, so as to obtain the target object behavior data corresponding to the at least one search result for explanation:
Assuming that at least one search result is a plurality of pictures, and the data form of each picture is < query-photo >, the calculation formula for correcting the ltr of each picture through the corresponding similarity can be as follows:
wherein sim is qp Refers to similarity, sim qp Alike refers to "sim qp * like ", i.e." sim qp Product of like).
In another specific embodiment, taking at least one search result as a picture, taking ftr as object behavior data corresponding to the at least one search result as an example, weighting the object behavior data corresponding to the at least one search result based on the weighting factor by taking similarity between the target search information and the at least one search result as the weighting factor, so as to obtain the target object behavior data corresponding to the at least one search result, and explaining the weighted object behavior data corresponding to the at least one search result.
Assuming that at least one search result is a plurality of pictures, and the data form of each picture is < query-photo >, the calculation formula for correcting the ftr of each picture through the corresponding similarity may be as follows:
wherein sim is qp Refers to similarity, sim qp The term "low" refers to "sim qp * Follow ", i.e." sim qp Product of the product with the following).
Assuming that at least one search result is 10 pictures, according to the similarity between the 10 pictures and the target search information, the object behavior data (including click rate data, praise rate data, attention rate data, etc.) corresponding to each of the 10 pictures can be modified to obtain the target object behavior data (including modified click rate data, praise rate data, attention rate data, etc.) corresponding to each of the 10 pictures.
In the embodiment of the disclosure, the similarity between the target search information and at least one search result is used as a weight factor, the object behavior data corresponding to at least one search result is weighted based on the weight factor, so that the target object behavior data corresponding to at least one search result is obtained by correcting the posterior object behavior data through the prior similarity, the situation that the posterior uncorrected object behavior data cannot accurately reflect the preference of the user under the condition that the returned result is an uncorrelated result is avoided, the deviation existing when the preference degree of the user is represented by the target object behavior data is reduced, and the accuracy of representing the preference of the user by the target object behavior data is improved.
In an alternative embodiment, FIG. 5 is a flow chart illustrating another behavioral data processing method according to an example embodiment. As shown in fig. 5, when the target object behavior data corresponding to the at least one search result is a preset number, the behavior data processing method may further include, after correcting the object behavior data corresponding to the at least one search result based on the similarity between the target search information and the at least one search result to obtain the target object behavior data corresponding to the at least one search result:
in step S19, based on the preset number of target object behavior data corresponding to the at least one search result, a ranking parameter corresponding to the at least one search result is determined.
In step S111, the at least one search result is ranked based on the ranking parameter corresponding to the at least one search result, so as to obtain a ranking sequence corresponding to the at least one search result.
In an optional embodiment, when at least one search result is a picture, determining, in the step S19, the ranking parameter corresponding to the at least one search result based on the preset number of target object behavior data corresponding to the at least one search result may include:
Multiplying the preset number of target object behavior data (including corrected click rate data, praise rate data, attention rate data and the like) corresponding to the at least one search result to obtain a ranking parameter corresponding to the at least one search result. The specific calculation formula can be as follows:
score qp ＝ctr qp *ltr qp *ftr qp ...，
wherein score qp Refers to the ordering parameter, qp refers to query-photo.
In an alternative embodiment, in step S111, each search result may be ranked based on the ranking parameter corresponding to each result, to obtain the ranking sequence corresponding to each search result.
Assuming that at least one search result is 10 pictures, score of the 10 pictures can be calculated in the manner of the above-mentioned ranking parameter calculation qp And follow the corresponding score for these 10 pictures qp And sequencing to obtain a sequencing sequence.
Because each target object behavior data (including corrected click rate data, praise rate data, attention rate data and the like) corresponding to each search result can reflect the preference of the user to the search result to a certain extent, a certain deviation may exist in the single target object behavior data when the preference of the user is described. Based on this, in the embodiment of the present disclosure, score obtained by multiplying a preset number of target object behavior data corresponding to each search result qp Based on the method, each search result is ranked, deviation of single target object behavior data in describing user preference can be effectively avoided, accuracy of determining ranking parameters corresponding to each search result is improved, and accuracy of recommending interested search results to users subsequently is improved.
In a particular embodiment, FIG. 6 is a flow chart illustrating another behavioral data processing method according to an example embodiment. As shown in fig. 6, the ranking of the at least one search result based on the ranking parameter corresponding to the at least one search result to obtain a ranking sequence corresponding to the at least one search result may include;
in step S1111, from the at least one search result, a search result having a similarity with the target search information less than or equal to a similarity threshold is filtered, and a filtered search result is obtained.
In step S1113, the filtered search results are ranked based on ranking parameters corresponding to the filtered search results, so as to obtain the ranking sequence.
Alternatively, in the step S1111, the similarity between each search result and the target search information may be compared, if the score of the search result qp If the similarity is larger than the corresponding similarity threshold, the search result is participated in the subsequent sorting process, and if the score of the search result is larger than the corresponding similarity threshold, the search result is selected to be sorted in the subsequent sorting process qp Less than or equal to the corresponding similarity threshold, then the score corresponding to the search result is considered qp Is 0, the ccore qp The search result of 0 does not participate in the subsequent ranking process, then the search result may be selected from the at least one search resultIn (3) the score qp Filtering the search result with the value of 0 to obtain a filtered search result.
Specifically, taking at least one search result as an example, a score corresponding to the search result qp The calculation formula of (2) can be as follows:
where θ is the similarity threshold.
Assuming that at least one search result is 10 pictures, through calculation according to the formula, it is found that the similarity between 7 search results and target search information is larger than a similarity threshold, the similarity between 3 search results and target search information is smaller than or equal to the similarity threshold, and the score corresponding to the 3 search results finally qp If 0, the score can be filtered from 10 pictures without participating in the subsequent sorting process qp And 3 search results of 0, and the 7 search results are filtered search results. Then, score corresponding to the 7 search results qp And re-ordering to obtain the ordered sequence, and recommending the search result in the ordered sequence to the user corresponding to the client.
In the embodiment of the disclosure, the similarity between each search result and the target search information is compared with the similarity threshold value, so that search results which are not subjected to parameter sorting are filtered, strong correlation among the search results which finally participate in sorting can be ensured, the problem that posterior data cannot accurately reflect the preference of a user under the condition that the returned search results are irrelevant search results is avoided, and the accuracy of describing the preference degree of the user is improved.
According to the behavior data processing method provided by the embodiment of the disclosure, since the corresponding object behavior data can be corrected through the similarity between the target search information and at least one search result (namely, the similarity determined through priori knowledge is corrected for the posterior object data behavior), the corrected target object behavior data can accurately reflect the preference of the user, and further, the deviation existing when the preference degree of the user is represented through the target object behavior data is reduced
FIG. 7 is a block diagram of a behavioural data processing apparatus, according to an example embodiment. Referring to fig. 7, the apparatus may include a search result acquisition module 31, a similarity determination module 32, an object behavior data determination module 33, and a correction module 34.
The search result acquisition module 31 is configured to perform acquisition of at least one search result corresponding to the target search information.
The similarity determination module 32 is configured to perform a determination of a similarity between the target search information and the at least one search result.
The object behavior data determining module 33 is configured to determine object behavior data corresponding to the at least one search result.
And a correction module 34 configured to perform correction on the object behavior data corresponding to the at least one search result based on the similarity between the target search information and the at least one search result, so as to obtain the target object behavior data corresponding to the at least one search result.
In an exemplary embodiment, the correction module 34 is configured to perform weighting processing on the object behavior data corresponding to the at least one search result based on the weighting factor by using the similarity between the target search information and the at least one search result as the weighting factor, so as to obtain the target object behavior data corresponding to the at least one search result.
In an exemplary embodiment, the target object behavior data corresponding to the at least one search result is a preset number, and the behavior data processing device may further include:
A ranking parameter determining module configured to perform determining a ranking parameter corresponding to the at least one search result based on a preset number of target object behavior data corresponding to the at least one search result;
and the sequencing sequence determining module is configured to execute sequencing of the at least one search result based on the sequencing parameter corresponding to the at least one search result to obtain a sequencing sequence corresponding to the at least one search result.
In an exemplary embodiment, the above-mentioned ordered sequence determining module may include:
a filtering unit configured to perform filtering of search results with a similarity with the target search information being less than or equal to a similarity threshold value from the at least one search result, to obtain filtered search results;
and the sorting unit is configured to execute sorting of the filtered search results based on sorting parameters corresponding to the filtered search results to obtain the sorting sequence.
In an exemplary embodiment, the similarity determining module 32 may include:
a first feature extraction unit configured to perform feature extraction on the target search information to obtain a first information feature extraction result corresponding to the target search information;
A second feature extraction unit configured to perform feature extraction on the at least one search result to obtain a second information feature extraction result corresponding to the at least one search result;
and a matching processing unit configured to perform similarity matching processing on the first information feature extraction result and the second information feature extraction result to obtain similarity between the target search information and the at least one search result, wherein the first information feature extraction result and the second information feature extraction result are in the same dimension.
In an exemplary embodiment, the matching processing unit is configured to perform similarity matching processing on the first information feature extraction result and the second information feature extraction result based on a similarity model, so as to obtain a similarity between the target search information and the at least one search result.
In an exemplary embodiment, the behavior data processing apparatus may further include:
a sample data set acquisition module configured to perform acquisition of a sample data set including at least one sample data having a similarity tag, the at least one sample data including sample search information and a sample search result corresponding to the sample search information, the similarity tag characterizing a similarity between the sample search information and the sample search result;
The first sample characteristic extraction module is configured to perform characteristic extraction on the sample search information to obtain a first sample information characteristic extraction result corresponding to the sample search information;
the second sample feature extraction module is configured to perform feature extraction on the sample search result to obtain a second sample information feature extraction result corresponding to the sample search result;
the training module is configured to perform similarity training on the neural network based on the first sample information feature extraction result and the second sample information feature extraction result to obtain a similarity prediction result between the first sample information feature extraction result and the second sample information feature extraction result;
a loss data determining module configured to determine loss data based on the similarity prediction result and the similarity label;
and a similarity model configured to perform training of the neural network based on the loss data to obtain the similarity model.
In an exemplary embodiment, the object behavior data determining module 33 may further include:
a log data obtaining unit configured to obtain log data corresponding to the at least one search result;
And the statistics unit is configured to perform statistics on the log data corresponding to the at least one search result to obtain object behavior data corresponding to the at least one search result.
In an exemplary embodiment, the statistics unit may further include:
an information obtaining subunit, configured to obtain, from log data corresponding to the at least one search result, display information and corresponding operation information of the at least one search result within a preset time;
and an object behavior data determining subunit configured to determine object behavior data corresponding to the at least one search result based on the display information and the corresponding operation information of the at least one search result within the preset time.
The specific manner in which the various modules perform the operations in the apparatus of the above embodiments have been described in detail in connection with the embodiments of the method, and will not be described in detail herein.
In an exemplary embodiment, there is also provided an electronic device including a processor; a memory for storing processor-executable instructions; wherein the processor is configured to implement the steps of any one of the above embodiments as a data processing method when executing instructions stored on the memory.
The electronic device may be a terminal, a server or similar computing device, which is exemplified by a server, fig. 8 is a block diagram of an electronic device for behavioral data processing according to an exemplary embodiment, where the electronic device 40 may vary considerably in configuration or performance, and may include one or more central processing units (Central Processing Units, CPU) 41 (the central processing unit 41 may include, but is not limited to, a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 43 for storing data, one or more storage media 42 (e.g., one or more mass storage devices) storing applications 423 or data 422. Wherein the memory 43 and the storage medium 42 may be transitory or persistent. The program stored on the storage medium 42 may include one or more modules, each of which may include a series of instruction operations in the electronic device. Still further, the central processor 41 may be arranged to communicate with a storage medium 42, executing a series of instruction operations in the storage medium 42 on the electronic device 40. The electronic device 40 may also include one or more power supplies 36, one or more wired or wireless network interfaces 45, one or more input/output interfaces 44, and/or one or more operating systems 421, such as Windows ServerTM, mac OS XTM, unixTM, linuxTM, freeBSDTM, etc.
The input-output interface 44 may be used to receive or transmit data via a network. Specific examples of the network described above may include a wireless network provided by a communications provider of electronic device 40. In one example, the input-output interface 44 includes a network adapter (Network Interface Controller, NIC) that can connect to other network devices through a base station to communicate with the internet. In an exemplary embodiment, the input/output interface 44 may be a Radio Frequency (RF) module for communicating with the internet wirelessly.
It will be appreciated by those of ordinary skill in the art that the configuration shown in fig. 8 is merely illustrative and is not intended to limit the configuration of the electronic device described above. For example, electronic device 40 may also include more or fewer components than shown in FIG. 8, or have a different configuration than shown in FIG. 8.
In an exemplary embodiment, a computer readable storage medium is also provided, which when executed by a processor of an electronic device, enables the electronic device to perform the steps of any one of the above embodiments as a data processing method.
In an exemplary embodiment, there is also provided a computer program product comprising a computer program which, when executed by a processor, implements the behavioural data processing method provided in any one of the embodiments described above.
Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in embodiments provided by the present disclosure may include non-volatile and/or volatile memory. The nonvolatile memory can include Read Only Memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link DRAM (SLDRAM), memory bus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), among others.
Other embodiments of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure disclosed herein. This disclosure is intended to cover any adaptations, uses, or adaptations of the disclosure following the general principles of the disclosure and including such departures from the present disclosure as come within known or customary practice within the art to which the disclosure pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.
It is to be understood that the present disclosure is not limited to the precise arrangements and instrumentalities shown in the drawings, and that various modifications and changes may be effected without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims (14)

1. A behavioural data processing method, comprising:
acquiring at least one search result corresponding to the target search information;
determining a similarity between the target search information and the at least one search result;
determining object behavior data corresponding to the at least one search result according to the ratio of the operation information and the display information of the at least one search result in a preset time;
And taking the similarity between the target search information and the at least one search result as a weight factor, and carrying out weighting processing on the object behavior data corresponding to the at least one search result based on the weight factor to obtain the target object behavior data corresponding to the at least one search result.
2. The behavioral data processing method according to claim 1, wherein, if the target object behavioral data corresponding to the at least one search result is a preset number, after the similarity between the target search information and the at least one search result is used as a weight factor, weighting the object behavioral data corresponding to the at least one search result based on the weight factor, to obtain the target object behavioral data corresponding to the at least one search result, the behavioral data processing method further includes:
determining a ranking parameter corresponding to the at least one search result based on a preset number of target object behavior data corresponding to the at least one search result;
and ordering the at least one search result based on the ordering parameter corresponding to the at least one search result to obtain an ordering sequence corresponding to the at least one search result.
3. The behavioral data processing method according to claim 2, wherein said ranking the at least one search result based on the ranking parameter corresponding to the at least one search result, to obtain the ranking sequence corresponding to the at least one search result, comprises:
filtering search results with similarity smaller than or equal to a similarity threshold value from the at least one search result, and obtaining filtered search results;
and ordering the filtered search results based on ordering parameters corresponding to the filtered search results to obtain the ordering sequence.
4. The behavioral data processing method according to claim 1, wherein said determining a similarity between said target search information and said at least one search result comprises:
extracting features of the target search information to obtain a first information feature extraction result corresponding to the target search information;
performing feature extraction on the at least one search result to obtain a second information feature extraction result corresponding to the at least one search result;
and performing similarity matching processing on the first information feature extraction result and the second information feature extraction result to obtain the similarity between the target search information and the at least one search result, wherein the first information feature extraction result and the second information feature extraction result are in the same dimension.
5. The behavioral data processing method of claim 4 wherein said performing a similarity matching process on said first information feature extraction result and said second information feature extraction result to obtain a similarity between said target search information and said at least one search result comprises:
performing similarity matching processing on the first information feature extraction result and the second information feature extraction result based on a similarity model to obtain similarity between the target search information and the at least one search result;
the generation mode of the similarity model comprises the following steps:
obtaining a sample data set, wherein the sample data set comprises at least one sample data, the at least one sample data has a similarity label, the at least one sample data comprises sample search information and a sample search result corresponding to the sample search information, and the similarity label represents similarity between the sample search information and the sample search result;
extracting features of the sample searching information to obtain a first sample information feature extraction result corresponding to the sample searching information;
Extracting features of the sample search results to obtain second sample information feature extraction results corresponding to the sample search results;
performing similarity training on a neural network based on the first sample information feature extraction result and the second sample information feature extraction result to obtain a similarity prediction result between the first sample information feature extraction result and the second sample information feature extraction result;
determining loss data based on the similarity prediction result and the similarity label;
and training the neural network based on the loss data to obtain the similarity model.
6. The behavioral data processing method according to any one of claims 1 to 5, wherein the determining the object behavioral data corresponding to the at least one search result according to the ratio of the operation information and the presentation information of the at least one search result within a preset time includes:
acquiring log data corresponding to the at least one search result;
acquiring operation information and display information of at least one search result in preset time from log data corresponding to the at least one search result;
And determining object behavior data corresponding to the at least one search result based on the ratio of the operation information and the display information of the at least one search result in the preset time.
7. A behavioural data processing apparatus, comprising:
the search result acquisition module is configured to execute at least one search result corresponding to the acquisition target search information;
a similarity determination module configured to perform a determination of a similarity between the target search information and the at least one search result;
the object behavior data determining module is configured to execute the determination of the object behavior data corresponding to the at least one search result according to the ratio of the operation information and the display information of the at least one search result in the preset time;
and the correction module is configured to perform weighting processing on the object behavior data corresponding to the at least one search result based on the weight factor by taking the similarity between the target search information and the at least one search result as the weight factor, so as to obtain the target object behavior data corresponding to the at least one search result.
8. The behavior data processing device according to claim 7, wherein the target object behavior data corresponding to the at least one search result is a preset number, and the behavior data processing device further comprises:
A ranking parameter determining module configured to perform determining a ranking parameter corresponding to the at least one search result based on a preset number of target object behavior data corresponding to the at least one search result;
and the ordering sequence determining module is configured to execute ordering of the at least one search result based on the ordering parameter corresponding to the at least one search result, so as to obtain an ordering sequence corresponding to the at least one search result.
9. The behavioral data processing apparatus of claim 8, wherein the ordered sequence determination module comprises:
a filtering unit configured to perform filtering of search results with a similarity with the target search information being less than or equal to a similarity threshold value from the at least one search result, to obtain filtered search results;
and the sorting unit is configured to sort the filtered search results based on the sorting parameters corresponding to the filtered search results to obtain the sorting sequence.
10. The behavioral data processing apparatus of claim 7, wherein the similarity determination module comprises:
the first feature extraction unit is configured to perform feature extraction on the target search information to obtain a first information feature extraction result corresponding to the target search information;
A second feature extraction unit configured to perform feature extraction on the at least one search result to obtain a second information feature extraction result corresponding to the at least one search result;
and the matching processing unit is configured to perform similarity matching processing on the first information feature extraction result and the second information feature extraction result to obtain similarity between the target search information and the at least one search result, wherein the first information feature extraction result and the second information feature extraction result are in the same dimension.
11. The behavioral data processing apparatus of claim 10,
the matching processing unit is configured to perform similarity matching processing on the first information feature extraction result and the second information feature extraction result based on a similarity model, so as to obtain similarity between the target search information and the at least one search result;
the behavior data processing apparatus further includes:
a sample data set acquisition module configured to perform acquiring a sample data set, the sample data set including at least one sample data, the at least one sample data having a similarity tag, the at least one sample data including sample search information and sample search results corresponding to the sample search information, the similarity tag characterizing a similarity between the sample search information and the sample search results;
The first sample characteristic extraction module is configured to perform characteristic extraction on the sample search information to obtain a first sample information characteristic extraction result corresponding to the sample search information;
the second sample feature extraction module is configured to perform feature extraction on the sample search result to obtain a second sample information feature extraction result corresponding to the sample search result;
the training module is configured to perform similarity training on the neural network based on the first sample information feature extraction result and the second sample information feature extraction result, and obtain a similarity prediction result between the first sample information feature extraction result and the second sample information feature extraction result;
a loss data determination module configured to perform determining loss data based on the similarity prediction result and the similarity label;
a similarity model determination module configured to perform training of the neural network based on the loss data to obtain the similarity model.
12. The behavior data processing apparatus according to any one of claims 7 to 11, wherein the object behavior data determination module includes:
A log data obtaining unit configured to obtain log data corresponding to the at least one search result;
an information obtaining subunit, configured to obtain operation information and display information of the at least one search result in a preset time from log data corresponding to the at least one search result;
and the object behavior data determining subunit is configured to determine object behavior data corresponding to the at least one search result based on the ratio of the operation information and the display information of the at least one search result in the preset time.
13. An electronic device, comprising:
a processor;
a memory for storing the processor-executable instructions; wherein the processor is configured to execute the instructions to implement the behavioural data processing method as claimed in any one of claims 1 to 6.
14. A computer readable storage medium, which when executed by a processor of an electronic device, causes the electronic device to perform the behavioural data processing method as claimed in any one of claims 1 to 6.
CN202110529224.1A 2021-05-14 2021-05-14 Behavior data processing method and device, electronic equipment and storage medium Active CN113420056B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110529224.1A CN113420056B (en) 2021-05-14 2021-05-14 Behavior data processing method and device, electronic equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110529224.1A CN113420056B (en) 2021-05-14 2021-05-14 Behavior data processing method and device, electronic equipment and storage medium

Publications (2)

Publication Number Publication Date
CN113420056A CN113420056A (en) 2021-09-21
CN113420056B true CN113420056B (en) 2023-12-26

Family

ID=77712423

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110529224.1A Active CN113420056B (en) 2021-05-14 2021-05-14 Behavior data processing method and device, electronic equipment and storage medium

Country Status (1)

Country Link
CN (1) CN113420056B (en)

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107153656A (en) * 2016-03-03 2017-09-12 阿里巴巴集团控股有限公司 A kind of information search method and device
CN109033140A (en) * 2018-06-08 2018-12-18 北京百度网讯科技有限公司 A kind of method, apparatus, equipment and the computer storage medium of determining search result
CN109086394A (en) * 2018-07-27 2018-12-25 天津字节跳动科技有限公司 Search ordering method, device, computer equipment and storage medium
CN109902224A (en) * 2019-01-17 2019-06-18 平安城市建设科技(深圳)有限公司 Source of houses recommended method, device, equipment and medium based on user behavior analysis
CN110162593A (en) * 2018-11-29 2019-08-23 腾讯科技（深圳）有限公司 A kind of processing of search result, similarity model training method and device
CN110532528A (en) * 2019-08-27 2019-12-03 掌阅科技股份有限公司 Books similarity calculating method and electronic equipment based on random walk
CN111400512A (en) * 2020-03-09 2020-07-10 北京达佳互联信息技术有限公司 Method and device for screening multimedia resources
CN111881343A (en) * 2020-07-07 2020-11-03 Oppo广东移动通信有限公司 Information pushing method and device, electronic equipment and computer readable storage medium

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107153656A (en) * 2016-03-03 2017-09-12 阿里巴巴集团控股有限公司 A kind of information search method and device
CN109033140A (en) * 2018-06-08 2018-12-18 北京百度网讯科技有限公司 A kind of method, apparatus, equipment and the computer storage medium of determining search result
CN109086394A (en) * 2018-07-27 2018-12-25 天津字节跳动科技有限公司 Search ordering method, device, computer equipment and storage medium
CN110162593A (en) * 2018-11-29 2019-08-23 腾讯科技（深圳）有限公司 A kind of processing of search result, similarity model training method and device
CN109902224A (en) * 2019-01-17 2019-06-18 平安城市建设科技(深圳)有限公司 Source of houses recommended method, device, equipment and medium based on user behavior analysis
CN110532528A (en) * 2019-08-27 2019-12-03 掌阅科技股份有限公司 Books similarity calculating method and electronic equipment based on random walk
CN111400512A (en) * 2020-03-09 2020-07-10 北京达佳互联信息技术有限公司 Method and device for screening multimedia resources
CN111881343A (en) * 2020-07-07 2020-11-03 Oppo广东移动通信有限公司 Information pushing method and device, electronic equipment and computer readable storage medium

Also Published As

Publication number Publication date
CN113420056A (en) 2021-09-21

Similar Documents

Publication Publication Date Title
CN109902708B (en) Recommendation model training method and related device
CN107862553B (en) Advertisement real-time recommendation method and device, terminal equipment and storage medium
CN107404656B (en) Live video recommended method, device and server
CN110019896B (en) Image retrieval method and device and electronic equipment
CN106326391B (en) Multimedia resource recommendation method and device
US20200349385A1 (en) Multimedia resource matching method and apparatus, storage medium, and electronic apparatus
CN106339507B (en) Streaming Media information push method and device
CN106528813A (en) Multimedia recommendation method and apparatus
CN107870984A (en) The method and apparatus for identifying the intention of search term
CN111651666A (en) User theme recommendation method and device, computer equipment and storage medium
CN113919923A (en) Live broadcast recommendation model training method, live broadcast recommendation method and related equipment
CN114245185B (en) Video recommendation method, model training method, device, electronic equipment and medium
CN113886711B (en) Content recommendation method, device, server and storage medium
CN114528474A (en) Method and device for determining recommended object, electronic equipment and storage medium
CN113407772B (en) Video recommendation model generation method, video recommendation method and device
CN112115354B (en) Information processing method, device, server and storage medium
CN113420056B (en) Behavior data processing method and device, electronic equipment and storage medium
CN111191065A (en) Homologous image determining method and device
CN113297417B (en) Video pushing method, device, electronic equipment and storage medium
CN114329055A (en) Search recommendation method and recommendation device, electronic device and storage medium
CN111597444B (en) Searching method, searching device, server and storage medium
CN114282115A (en) Content recommendation method
CN113901324A (en) Search object ordering method, device, equipment and storage medium
CN113434775A (en) Method and device for determining search content
CN114519593A (en) Resource recall model updating method and device, electronic equipment and storage medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant