CN112328889A

CN112328889A - Method and device for determining recommended search terms, readable medium and electronic equipment

Info

Publication number: CN112328889A
Application number: CN202011325311.7A
Authority: CN
Inventors: 彭睿棋
Original assignee: Beijing ByteDance Network Technology Co Ltd
Current assignee: Beijing ByteDance Network Technology Co Ltd
Priority date: 2020-11-23
Filing date: 2020-11-23
Publication date: 2021-02-05

Abstract

The disclosure relates to a method and a device for determining recommended search terms, a readable medium and electronic equipment. The method comprises the following steps: acquiring search associated information corresponding to the search terms; determining a plurality of associated words of the search word according to the search associated information; determining a target value parameter of each associated word, wherein the target value parameter is used for representing value information corresponding to the clicked associated word and value information corresponding to the displayed associated word; and determining a recommended search word from the associated words according to the target value parameter of the associated word. Therefore, by introducing the standard of the target value parameter, different types of associated words which cannot be directly compared originally have comparability, the original relative relationship is kept, different types of associated words can be taken into consideration, the diversity of the determination of recommended search words is improved, richer search words are recommended for the user, the search path of the user can be shortened, and the search requirement of the user can be better met.

Description

Method and device for determining recommended search terms, readable medium and electronic equipment

Technical Field

The present disclosure relates to the field of computer technologies, and in particular, to a method and an apparatus for determining recommended search terms, a readable medium, and an electronic device.

Background

With the development of internet technology, more and more information is transmitted through the internet, and with the increase of information amount, the demand of users for search functions is gradually increased, and more users acquire and query information through network search, for example, search through keywords. At present, when a user utilizes a search engine to search through keywords, some recommended search terms can be displayed for the user at the same time to provide guidance for the search of the user and provide more search possibilities.

In the related art, what kind of search terms are recommended usually depends on a single evaluation criterion, however, the kinds of candidate terms are various, the use of the single criterion is often only applicable to the evaluation of a certain candidate term, and other kinds of candidate terms may not be comparable under the criterion, so that other kinds of candidate terms are always ignored and not recommended under the criterion, resulting in a single finally recommended search term, which cannot help to shorten the search path of the user, and thus the search requirement of the user cannot be solved.

Disclosure of Invention

This summary is provided to introduce a selection of concepts in a simplified form that are further described below in the detailed description. This summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter.

In a first aspect, the present disclosure provides a method for determining recommended search terms, the method including:

acquiring search associated information corresponding to the search terms;

determining a plurality of associated words of the search word according to the search associated information;

determining a target value parameter of each associated word, wherein the target value parameter is used for representing value information corresponding to the clicked associated word and value information corresponding to the displayed associated word;

and determining a recommended search word from the associated words according to the target value parameter of the associated word.

In a second aspect, an apparatus for determining recommended search terms is provided, the apparatus comprising:

the acquisition module is used for acquiring search associated information corresponding to the search terms;

the first determining module is used for determining a plurality of associated words of the search word according to the search associated information;

the second determining module is used for determining a target value parameter of each associated word, wherein the target value parameter is used for representing value information corresponding to the clicked associated word and value information corresponding to the displayed associated word;

and the third determining module is used for determining the recommended search word from the associated words according to the target value parameter of the associated words.

In a third aspect, a computer-readable medium is provided, on which a computer program is stored, which program, when being executed by a processing device, carries out the steps of the method according to the first aspect of the disclosure.

In a fourth aspect, an electronic device is provided, comprising:

a storage device having a computer program stored thereon;

processing means for executing the computer program in the storage means to implement the steps of the method of the first aspect of the present disclosure.

According to the technical scheme, the search associated information corresponding to the search word is obtained, the multiple associated words of the search word are determined according to the search associated information, the target value parameter of each associated word is determined, and then the recommended search word is determined from the multiple associated words according to the target value parameter of the associated word. The target value parameter is used for representing value information corresponding to the clicked related word and value information corresponding to the displayed related word. Therefore, when the recommended search terms are determined based on the relevant terms, by introducing the standard of the target value parameter, different types of relevant terms which cannot be directly compared through a single standard originally have comparability, and the original relative relation cannot be influenced, so that different types of relevant terms can be taken into consideration, the diversity of determination of the recommended search terms is improved, richer search terms are recommended for the user, the search path of the user is favorably shortened, the search requirements of the user are better met, and the user experience is improved. In addition, under the condition that the recommended search terms can help the platform to obtain the reward, based on the various recommended search terms, the profit can be further created for the platform in a richer mode.

Additional features and advantages of the disclosure will be set forth in the detailed description which follows.

Drawings

The above and other features, advantages and aspects of various embodiments of the present disclosure will become more apparent by referring to the following detailed description when taken in conjunction with the accompanying drawings. Throughout the drawings, the same or similar reference numbers refer to the same or similar elements. It should be understood that the drawings are schematic and that elements and features are not necessarily drawn to scale. In the drawings:

fig. 1 is a flowchart of a recommended search term determination method provided according to an embodiment of the present disclosure;

fig. 2 is an exemplary flowchart of a step of determining a plurality of related words of the search term according to the search related information in the recommended search term determining method provided by the present disclosure;

fig. 3 is an exemplary flowchart for determining a target value parameter of a related word in the recommended search word determining method according to the present disclosure;

FIG. 4 is a block diagram of a recommended search term determination apparatus provided in accordance with one embodiment of the present disclosure;

FIG. 5 illustrates a schematic diagram of an electronic device suitable for use in implementing embodiments of the present disclosure.

Detailed Description

Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While certain embodiments of the present disclosure are shown in the drawings, it is to be understood that the present disclosure may be embodied in various forms and should not be construed as limited to the embodiments set forth herein, but rather are provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the disclosure are for illustration purposes only and are not intended to limit the scope of the disclosure.

It should be understood that the various steps recited in the method embodiments of the present disclosure may be performed in a different order, and/or performed in parallel. Moreover, method embodiments may include additional steps and/or omit performing the illustrated steps. The scope of the present disclosure is not limited in this respect.

The term "include" and variations thereof as used herein are open-ended, i.e., "including but not limited to". The term "based on" is "based, at least in part, on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Relevant definitions for other terms will be given in the following description.

It should be noted that the terms "first", "second", and the like in the present disclosure are only used for distinguishing different devices, modules or units, and are not used for limiting the order or interdependence relationship of the functions performed by the devices, modules or units.

It is noted that references to "a", "an", and "the" modifications in this disclosure are intended to be illustrative rather than limiting, and that those skilled in the art will recognize that "one or more" may be used unless the context clearly dictates otherwise.

The names of messages or information exchanged between devices in the embodiments of the present disclosure are for illustrative purposes only, and are not intended to limit the scope of the messages or information.

As described in the background art, as the amount of internet information increases, more and more users acquire and query information by means of web search. Among them, the most common way is by keyword search. At present, when a user searches through keywords, some recommended search terms are generally displayed for the user at the same time, so that search guidance is provided for the user, and more search possibilities are provided for the user.

In the related art, displaying which recommended search word generally depends on sorting some candidate words that can be used as the recommended search word, that is, sorting some candidate words with a predetermined standard first, and then displaying some candidate words with the top sorting as the recommended search word to the user. In this process, however, the criteria relied upon are generally singular. For example, there are some candidate words with commercial values, which may help the platform to obtain a certain degree of revenue return, each candidate word may correspond to a proposed value (the proposed value may indicate a value that the candidate word needs to be clicked or viewed once after being displayed), when selecting a recommended search word from the candidate words, a product is generally calculated according to the proposed value of each candidate word and a click rate of the candidate word itself, then the candidate words are ranked according to the product, and a finally recommended search word is determined according to a ranking result. However, these alternatives generally include two alternatives, one is a first kind of words that can only get a reward when clicked, and the other is a second kind of words that can get a reward when displayed, because these two kinds of words get a reward in different ways, the respective determination criteria of the drawn value also have a large difference, and the two are obviously not comparable, but belong to a single criterion for the first kind of words by way of the product of the drawn value and the click rate, therefore, in the ranking result, the second kind of words rarely appear at the front position in the ranking result, and naturally also rarely are displayed as recommended search words. Therefore, the situation that one kind of alternative words are ranked more forward and other kinds of alternative words are ranked more backward easily caused by using a single standard can lead to the fact that the kind of the finally recommended search words is single, more various search words can not be recommended for the user, and the search requirement of the user can not be better met.

In order to solve the above problems, the present disclosure provides a method, an apparatus, a readable medium and an electronic device for determining recommended search terms, so as to improve the diversity of determining recommended search terms and better meet the search requirements of users.

Fig. 1 is a flowchart of a recommended search term determination method provided according to an embodiment of the present disclosure, which may include the following steps, as shown in fig. 1.

In step 11, search related information corresponding to the search term is acquired.

The search related information corresponding to the search term may be obtained by collecting search records related to the search term in each platform (or, a designated platform). The search related information may be any information related to the search term, where the search term may be input by the user in real time or obtained based on the historical search behavior of the user (for example, what contents the user has searched once, all or part of the contents may be used as the search term). For example, the search related information corresponding to the search term may include, but is not limited to, at least one of the following: search result information corresponding to the search terms, interaction behavior information associated with the search terms, and search behavior information associated with the search terms.

The search result information corresponding to the search term may be obtained from search results obtained after searching for the search term all the time. For example, the search result information corresponding to the search term may include, but is not limited to, at least one of the following: the title of the search result, the web page information corresponding to the web page indicated by the search result, and the like.

The interactive behavior information associated with the search term can be obtained through search results obtained after different users search the search term all the time and operations of the users on the search results. Illustratively, the interaction behavior information may include, but is not limited to, at least one of: which search results are presented to the user, which search results are clicked by the user, the length of time the user dwells on the page indicated by the search results after clicking on the search results, and the like.

Search behavior information associated with a search term may be obtained from search behaviors of different users (e.g., all users or a specified user) searching for the search term. Illustratively, the search behavior information may include, but is not limited to, at least one of: all search terms searched by different users searching for the search term, search terms searched by different users searching for the search term within a specified time period before (and/or after) the search term, search terms searched by different users searching for the search term within a specified number of searches before (and/or after) the search term, and the like.

As another example, in addition to the examples given above, the search related information corresponding to a search term may include other terms that are similar and/or semantically similar to the present search term text.

In step 12, a plurality of related words of the search word are determined based on the search related information.

After the search related information is obtained, a plurality of related words of the search word are determined according to the obtained search related information. When determining the related word, a determination standard of the related word may be constructed in advance, and the related word corresponding to the search word is determined based on the search related information according to the established determination standard.

In step 13, a target value parameter of each related word is determined.

The target value parameter is used for representing value information corresponding to the clicked related word and value information corresponding to the displayed related word.

That is, by constructing a target value parameter in combination with both the value information corresponding to the clicked related word and the value information corresponding to the presented related word, and using the target value parameter as a unified evaluation criterion for the related word, it is possible to normalize the value of the related word, and thus, the target value parameter of the related word can reflect both the value corresponding to the clicked related word and the value corresponding to the presented related word. Furthermore, by using the unified standard of the target value parameter, different associated words can be compared with each other under the standard, and the method has comparability, so that the condition that a certain word is not recommended because the word cannot be compared does not occur any more.

In step 14, a recommended search term is determined from the plurality of related terms based on the target value parameter of the related term.

After the target value parameter of each related word is determined through step 13, the recommended search word can be determined from the plurality of related words according to the target value parameter of the related word. Illustratively, according to the target value parameter of each relevant word, the relevant words can be filtered through a specified rule to determine the recommended search word.

As described above, each related word uses the unified criterion of the target value parameter, so that each related word has comparability, and further, the related words can be evaluated based on the criterion to select the recommended search word.

In order to make those skilled in the art understand the technical solutions provided by the embodiments of the present invention, the following describes the corresponding steps in more detail.

First, a detailed description will be given of the case where a plurality of related words of the search word are specified based on the search related information in step 12. As described above, when determining a related word, a determination criterion of the related word may be constructed in advance, and the related word corresponding to the search word may be determined based on the search related information according to the constructed determination criterion.

In one possible implementation, the selected criteria for the associated word may be constructed based on requirements for text similarity (and/or semantic similarity). For example, if the pre-constructed determination criterion of the relevant word is that the text similarity is greater than a first threshold and the semantic similarity is greater than a second threshold, after the search relevant information is obtained, the search relevant information may be first subjected to word segmentation processing to obtain a plurality of segmented words, the text similarity and the semantic similarity between each segmented word and the search word are calculated, and then the segmented words whose text similarity is greater than the first threshold and the semantic similarity is greater than the second threshold are determined as the relevant words. In this way, a plurality of related words related to the search word can be quickly determined based on similarity of text or semantics.

In another possible implementation, the selected criteria for the relevant words may be constructed based on the requirements for the search behavior. For example, if the pre-constructed determination criterion of the related words is the content searched by the user within the specified time period before searching the search word, after obtaining the search related information, all the contents searched by the search word within the specified time period before being searched may be first located, and the related words may be determined based on the contents (for example, some words are extracted from the contents as related words). In this way, a plurality of related words related to the search word can be quickly determined based on the search behavior related to the search word.

In addition, the search related information may include a plurality of items of related information. Taking the example given above regarding the search related information as an example, assuming that the search related information includes search result information corresponding to a search term, and the search result information includes a title, then one title in the search result is one related information, and if there are multiple titles in the search result, the multiple titles correspond to multiple related information. Accordingly, in another possible embodiment, step 12 may include the following steps, as shown in fig. 2:

in step 21, determining the relevance between each piece of relevant information and the search word based on the relevance model;

in step 22, determining target associated information with the association degree greater than an association degree threshold value in the plurality of items of associated information;

in step 23, performing word segmentation processing on the target associated information, and determining the importance corresponding to each word segmentation;

in step 24, determining a core word of the target associated information according to the corresponding importance of each word segmentation;

in step 25, a plurality of related words of the search word are acquired based on the core word.

Based on step 21, for each item of associated information, a degree of association between each item of associated information and the search term may be determined based on the degree of association model. Wherein the degree of association may reflect a correlation between the two.

For example, the relevance information of the relevance to be determined may be input to a pre-trained relevance model, and an output result of the relevance model for the relevance information may be obtained as the relevance between the relevance information and the search term.

Wherein, the relevance model can be obtained by training a plurality of groups of training data. Each set of training data comprises historical association information and historical association degree corresponding to the historical association information. The meaning and obtaining manner of the historical associated information and the associated information are the same, and are not described herein again. The historical association degree corresponding to the historical association information can be obtained through the interaction between the user and the historical association information, wherein the interaction can be, for example, clicking, selecting, agreeing, and the like. For example, if the history related information is displayed to the user but not clicked by the user, the history related degree corresponding to the history related information may be recorded as 0, and if the history related information is displayed to the user and clicked by the user, the history related degree corresponding to the history related information may be recorded as 1. When the model is trained, a plurality of groups of training data are used for training, in each training process, historical association information in one group of training data is used as input data, historical association degree of the group of training data is used as target output of the model, a loss value is calculated according to the target output and the actual output by combining actual output of the model and is used for adjusting internal parameters of the model, and the trained model is used as an association degree model after the condition that the model stops training is met. The output of the relevance model is a numerical value between 0 and 1, and the larger the numerical value output by the relevance model is, the stronger the correlation between the relevance information input into the model and the search word is. It should be noted that the specific manner of model training is well known in the art, and the above words are only exemplary and are not intended to limit the disclosure.

After determining the degree of association between each item of associated information and the search term, in step 22, the target associated information having a degree of association greater than the threshold degree of association is determined among the plurality of items of associated information. As described above, the greater the degree of association between the associated information and the search word, the stronger the correlation between the associated information and the search word is. Therefore, in this way, the related information having a strong correlation with the search term can be specified from the plurality of items of related information as the target related information.

Since the target related information is directly extracted from the search related information, redundant or unnecessary information may exist, after the target related information is determined, a representative core word may be further extracted from the target related information. Therefore, the core word can be determined through

steps

23 and 24.

In step 23, the target related information is subjected to word segmentation processing, and the importance degree corresponding to each word segmentation is determined.

In step 24, the core word of the target associated information is determined according to the corresponding importance of each word segmentation.

The word segmentation processing is performed on the target associated information, and a currently common word segmentation method can be adopted, which is not described herein again.

The importance is the product of the word frequency of the participle appearing in the target associated information and the inverse document frequency of the participle in the corpus. That is, the importance of the participle can be represented by a TF-IDF (Term Frequency-Inverse text Frequency index) value, and the higher the TF-IDF value is, the more important the participle is for the text. The main idea of TF-IDF is that if a word occurs less frequently in a corpus and more frequently in a text, the more important the word is for the text, the more representative the text, and thus, the word can be used as a core word of the text. Based on the idea, the importance degree corresponding to each participle can be determined according to the calculation mode, and the participle with higher importance degree is determined as the core word of the target associated information.

After determining the core word of the target related information, step 25 may be performed to obtain a plurality of related words of the search word based on the core word.

In one possible implementation, the core word may be directly determined as the related word.

In another possible embodiment, step 25 may include the steps of:

acquiring the associated information of the core words;

performing word segmentation processing on the associated information of the core words to obtain a plurality of core words;

determining the relevance between each core participle and the core word based on the relevance model;

determining the core word segmentation corresponding to the maximum first L relevance degrees as a plurality of candidate words having relevance with the core word;

and filtering the plurality of candidate words, and taking the candidate words contained in the filtering result as the relevant words of the search words.

The related information of the core word may include at least one of search result information corresponding to the core word, interaction behavior information related to the core word, and search behavior information related to the core word. The search result information corresponding to the core word, the interaction behavior information associated with the core word, and the specific content of the search behavior information associated with the core word may refer to the search result information, the interaction behavior information associated with the search word, and the search behavior information associated with the search word in sequence, which are not described herein again.

The associated information of the core words is subjected to word segmentation processing to obtain a plurality of core words, and a currently common word segmentation method can be adopted, which is not described herein any more.

After obtaining a plurality of core participles, based on the relevance model, inputting the core participles with the relevance to be determined into the pre-trained relevance model, and obtaining an output result of the relevance model for the core participles as the relevance between the core participles and the core words.

After determining the association degrees between the core participles and the core words, the core participles corresponding to the largest first L association degrees may be determined as a plurality of candidate words having a correlation with the core words, where L is a positive integer.

Further, filtering processing may be performed on the plurality of candidate words, and the candidate words included in the filtering result may be used as the relevant words of the search word.

For example, the interaction probabilities of the candidate words included in the filtering result may be all higher than a first preset threshold. The interaction probability of the candidate words can be obtained by using a pre-trained pre-estimation model. The pre-estimation model can predict the interaction probability of the specified words. Illustratively, the interaction probability may be a Click Through Rate, and accordingly, the predictive model may use a CTR (Click-Through-Rate) predictive model. Data and model training modes required by model training of the CTR estimation model belong to common knowledge in the field and are not listed here.

Therefore, the candidate words with lower interaction probability can be filtered, the situation that the candidate words become recommended search words and cause resource waste is avoided, the candidate words in the filtering result have higher interaction probability, and the searching requirements of users can be met better.

For another example, if there are multiple segmented words with text similarity (and/or semantic similarity) higher than the second preset threshold in the candidate words, one or more of the multiple segmented words may be included in the filtering result. That is, the deduplication processing may be performed based on the text similarity or the semantic similarity, that is, only one or a few of the multiple segmented words whose texts are similar or whose semantics are similar may be retained. Therefore, the effect of removing duplication can be achieved, similar words in the associated words are avoided, and the quality of the associated words is improved.

According to the method, the target associated information with strong relevance to the search word is determined through the relevance, the core word is extracted from the target associated information, and the associated word is determined based on the core word, so that the associated word of the search word can be comprehensively and accurately determined in a layer-by-layer screening mode, accurate data support is provided for subsequent recommended search words, meanwhile, unnecessary information is gradually removed in a layer-by-layer screening mode, the data processing amount can be reduced, and subsequent unnecessary data processing is avoided.

The following describes in detail the target value parameter for determining a related word in step 13.

In one possible implementation, the target value parameter of the related word may be determined by the following method, as shown in fig. 3:

in step 31, click value information of the associated word is acquired;

in step 32, determining a first value parameter of the associated word according to the click value information;

in step 33, obtaining the presentation value information of the associated word, the interactive data of the associated word in the target history time period and the delivery return rate of the associated word in the target history time period;

in step 34, determining a second value parameter of the associated word according to the display value information, the interactive data and the delivery return rate;

in step 35, a target value parameter of the related word is determined based on the first value parameter and the second value parameter.

The first value parameter is used for representing value information corresponding to the clicked related word, and the second value parameter is used for representing value information corresponding to the displayed related word.

The click value information may be a single click value, that is, a value that the associated word is spent (or obtained) by a single click. Wherein the single click value is a numerical value that can be directly obtained based on the related word. Illustratively, the click value information may be a single click value. As another example, the click value information may be a value of P clicks, where P is a positive integer.

Illustratively, click value information may be determined directly as the first value parameter.

For another example, a product of the click value information and a first preset coefficient may be determined as the first value parameter, wherein the first preset coefficient may be set according to an empirical value.

For another example, the product of the single click value of the related word and the interaction probability of the related word may be used as the first value parameter. Wherein the interaction probability can be predicted by the prediction model as described above.

The presentation value information may be a single presentation value, that is, a value spent (or obtained) when the related word is presented once. Wherein, the single presentation value is a numerical value which can be directly obtained based on the relevant word. The display value information can also be the display value for M times, wherein M is a positive integer.

In a possible implementation manner, if the presentation value information is a single presentation value, determining the second value parameter of the associated word according to the presentation value information, the interaction data and the delivery return rate may include the following steps:

and determining the second value parameter according to the product of the single showing value, the interaction data and the delivery return rate.

For example, the product of the single exposure value, the interaction data and the delivery return rate may be determined as the second value parameter. For another example, a product of the single-show value, the interaction data, the delivery return rate, and a second preset coefficient may be determined as the second price parameter, where the second preset coefficient may be set according to an empirical value.

After determining the first value parameter and the second value parameter, a target value parameter of the related word may be determined. For example, the sum of the first value parameter and the second value parameter may be used as the target value parameter. For another example, a product of a sum of the first value parameter and the second value parameter and a third preset coefficient may be used as the target value parameter, where the third preset coefficient may be set according to an empirical value.

Therefore, the target value parameter of each relevant word can be determined through the method. It should be noted that the target value parameter may adopt various calculation methods according to actual requirements, and it is only necessary to ensure that the calculation methods adopted by each relevant word are the same.

Through the mode, the target value parameter is determined by combining the click value information and the display value information of the associated word, so that the value characteristic of the associated word can be more comprehensively reflected by the target value parameter, and the recommended search word can be conveniently determined by taking the target value parameter as an evaluation standard in the subsequent determination.

Returning now to fig. 1, in step 14, determining a recommended search term from a plurality of related terms according to the target value parameter of the related term may include the following steps:

determining a recommendation value corresponding to the associated word according to the target value parameter of the associated word and the interaction probability of the associated word;

and determining the relevant words corresponding to the maximum top N recommendation values as the recommended search words according to the recommendation values corresponding to the relevant words. Wherein N is a positive integer.

Wherein, the interaction probability of the associated word can be predicted by the prediction model as described above. For example, the interaction probability of the associated word may be determined as follows:

acquiring related words having correlation with the related words and interaction probability of each related word;

and inputting the related words and the interaction probability of each related word into a pre-trained pre-estimation model to obtain an output result of the pre-estimation model as the interaction probability of the related words.

For example, the related word having a correlation with the related word may be obtained by:

acquiring the associated information of the associated word;

performing word segmentation processing on the associated information of the associated words to obtain a plurality of associated words;

determining the relevance between each relevant participle and the relevant word based on the relevance model;

and determining the related participle corresponding to the maximum first K relevance degrees as the related word having the relevance with the related word. Wherein K is a positive integer.

The related information of the related word may include at least one of search result information corresponding to the related word, interaction behavior information associated with the related word, and search behavior information associated with the related word. The specific contents of the search result information corresponding to the related word, the interaction behavior information associated with the related word, and the search behavior information associated with the related word may refer to the search result information, the interaction behavior information associated with the search word, and the search behavior information associated with the search word in sequence, which are not described herein again.

The associated information of the associated word is segmented to obtain a plurality of associated words, and a currently common segmentation method can be adopted, which is not described herein.

After obtaining a plurality of associated participles, inputting the associated participles with the association degree to be determined to a pre-trained association degree model based on the association degree model, and obtaining an output result of the association degree model aiming at the associated participles as the association degree between the associated participles and the associated words.

After the association degrees between the associated participles and the associated words are determined, the associated participle corresponding to the top K maximum association degrees can be determined as the associated word having the correlation with the associated word.

After the related words having the relevance with the related words are obtained in the above manner, the interaction probability of each related word can be directly determined, and the interaction probability can be directly obtained according to the related words. Furthermore, the related words and the interaction probability of each related order can be input into the estimation model, and the output result of the estimation model is obtained as the interaction probability of the related words. If the interaction probability is the click rate, the pre-estimated model can be a CTR pre-estimated model.

Illustratively, the predictive model may be obtained by: acquiring training related words having correlation with the training words and the interaction probability of each training related word; inputting the training related words into the neural network model to obtain the actual output of the neural network model, taking the interaction probability of the training related words input into the neural network model as the target output of the neural network model, calculating a loss value according to the actual output and the target output of the neural network model to adjust the internal parameters of the model until the training stopping condition is met, and taking the trained model as an estimation model.

The obtaining mode of the training related word having a correlation with the training word is the same as the obtaining mode of the related word having a correlation with the related word in principle, and is not described herein again. And the interaction probability of the training related words can be directly obtained.

After the target value parameter of the associated word and the interaction probability of the associated word are obtained, the recommendation value corresponding to the associated word can be determined based on the target value parameter and the interaction probability of the associated word. For example, the product of the target value parameter of the related word and the interaction probability of the related word may be determined as the recommendation value corresponding to the related word.

Therefore, according to the recommendation value corresponding to each relevant word, the relevant word corresponding to the top N maximum recommendation values can be determined as the recommended search word.

After determining the recommended search terms, the recommended search terms may be used for presentation to the user so that the user can interact, e.g., browse, click, like, etc. For example, the presentation order of the recommended search terms may refer to the recommendation values corresponding to the related terms, for example, the recommendation values are arranged from left to right (or from top to bottom, etc.) in descending order.

Fig. 4 is a block diagram of a recommended search term determination apparatus provided according to an embodiment of the present disclosure, and as shown in fig. 4, the apparatus 40 may include:

an obtaining module 41, configured to obtain search related information corresponding to a search term;

a first determining module 42, configured to determine, according to the search related information, a plurality of related words of the search word;

a second determining module 43, configured to determine a target value parameter of each relevant word, where the target value parameter is used to represent value information corresponding to the clicked relevant word and value information corresponding to the displayed relevant word;

and the third determining module 44 is configured to determine, according to the target value parameter of the associated word, a recommended search word from the associated words.

Optionally, the search related information includes a plurality of items of related information;

the first determination module 42 includes:

the first determining submodule is used for determining the association degree between each piece of association information and the search word based on an association degree model;

the second determining submodule is used for determining target associated information of which the association degree is greater than an association degree threshold value in the plurality of items of associated information;

a third determining sub-module, configured to perform word segmentation processing on the target associated information, and determine an importance degree corresponding to each word, where the importance degree is a product of a word frequency of the word appearing in the target associated information and an inverse document frequency of the word in the corpus;

the fourth determining submodule is used for determining the core words of the target associated information according to the corresponding importance of each participle;

and the first obtaining sub-module is used for obtaining a plurality of associated words of the search word based on the core word.

Optionally, the first obtaining sub-module is configured to:

acquiring relevant information of a core word, wherein the relevant information of the core word comprises at least one of search result information corresponding to the core word, interactive behavior information associated with the core word and search behavior information associated with the core word;

determining the relevance between each core participle and the core word based on a relevance model;

determining core participles corresponding to the maximum first L relevance degrees as a plurality of candidate words having relevance with the core words, wherein L is a positive integer;

filtering the candidate words, and taking the candidate words contained in the filtering result as the relevant words of the search words;

and the interaction probability of each candidate word contained in the filtering result is higher than a first preset threshold.

Optionally, the second determining module 43 determines the target value parameter of the related word through the following sub-modules:

the second obtaining submodule is used for obtaining click value information of the associated word;

a fifth determining submodule, configured to determine a first value parameter of the associated word according to the click value information, where the first value parameter is used to represent value information corresponding to the clicked associated word;

the third obtaining submodule is used for obtaining the display value information of the associated word, the interactive data of the associated word in the target historical time period and the putting return rate of the associated word in the target historical time period;

a sixth determining submodule, configured to determine a second value parameter of the associated word according to the presentation value information, the interaction data, and the delivery return rate, where the second value parameter is used to represent value information corresponding to the presented associated word;

and the seventh determining submodule is used for determining the target value parameter of the associated word according to the first value parameter and the second value parameter.

Optionally, the display value information is a single display value;

the sixth determination submodule is configured to:

and determining the second price parameter according to the product of the single showing value, the interaction data and the delivery return rate.

Optionally, the third determining module 44 includes:

the eighth determining submodule is used for determining a recommendation value corresponding to the associated word according to the target value parameter of the associated word and the interaction probability of the associated word;

and the ninth determining submodule is used for determining the relevant word corresponding to the maximum first N recommendation values as the recommended search word according to the recommendation value corresponding to the relevant word, wherein N is a positive integer.

Optionally, the interaction probability of the related word is determined by:

obtaining related words having relevance with the related words and interaction probability of each related word;

and inputting the related words and the interaction probability of each related word into a pre-trained pre-estimation model, and obtaining an output result of the pre-estimation model as the interaction probability of the related words.

Optionally, the related word having a correlation with the related word is obtained by:

acquiring relevant information of a relevant word, wherein the relevant information of the relevant word comprises at least one of search result information corresponding to the relevant word, interactive behavior information associated with the relevant word and search behavior information associated with the relevant word;

determining the association degree between each associated participle and the associated word based on an association degree model;

and determining the related participles corresponding to the maximum first K relevance degrees as related words having relevance to the related words, wherein K is a positive integer.

Optionally, the search related information includes at least one of: the search result information corresponding to the search word, the interaction behavior information associated with the search word, and the search behavior information associated with the search word.

With regard to the apparatus in the above-described embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment related to the method, and will not be elaborated here.

Referring now to FIG. 5, a block diagram of an electronic device 600 suitable for use in implementing embodiments of the present disclosure is shown. The terminal device in the embodiments of the present disclosure may include, but is not limited to, a mobile terminal such as a mobile phone, a notebook computer, a digital broadcast receiver, a PDA (personal digital assistant), a PAD (tablet computer), a PMP (portable multimedia player), a vehicle terminal (e.g., a car navigation terminal), and the like, and a stationary terminal such as a digital TV, a desktop computer, and the like. The electronic device shown in fig. 5 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiments of the present disclosure.

As shown in fig. 5, electronic device 600 may include a processing means (e.g., central processing unit, graphics processor, etc.) 601 that may perform various appropriate actions and processes in accordance with a program stored in a Read Only Memory (ROM)602 or a program loaded from a storage means 608 into a Random Access Memory (RAM) 603. In the RAM 603, various programs and data necessary for the operation of the electronic apparatus 600 are also stored. The processing device 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input/output (I/O) interface 605 is also connected to bus 604.

Generally, the following devices may be connected to the I/O interface 605: input devices 606 including, for example, a touch screen, touch pad, keyboard, mouse, camera, microphone, accelerometer, gyroscope, etc.; output devices 607 including, for example, a Liquid Crystal Display (LCD), a speaker, a vibrator, and the like; storage 608 including, for example, tape, hard disk, etc.; and a communication device 609. The communication means 609 may allow the electronic device 600 to communicate with other devices wirelessly or by wire to exchange data. While fig. 5 illustrates an electronic device 600 having various means, it is to be understood that not all illustrated means are required to be implemented or provided. More or fewer devices may alternatively be implemented or provided.

In particular, according to an embodiment of the present disclosure, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program carried on a non-transitory computer readable medium, the computer program containing program code for performing the method illustrated by the flow chart. In such an embodiment, the computer program may be downloaded and installed from a network via the communication means 609, or may be installed from the storage means 608, or may be installed from the ROM 602. The computer program, when executed by the processing device 601, performs the above-described functions defined in the methods of the embodiments of the present disclosure.

It should be noted that the computer readable medium in the present disclosure can be a computer readable signal medium or a computer readable storage medium or any combination of the two. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples of the computer readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present disclosure, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In contrast, in the present disclosure, a computer readable signal medium may comprise a propagated data signal with computer readable program code embodied therein, either in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: electrical wires, optical cables, RF (radio frequency), etc., or any suitable combination of the foregoing.

In some embodiments, the clients, servers may communicate using any currently known or future developed network Protocol, such as HTTP (HyperText Transfer Protocol), and may interconnect with any form or medium of digital data communication (e.g., a communications network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), the Internet (e.g., the Internet), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future developed network.

The computer readable medium may be embodied in the electronic device; or may exist separately without being assembled into the electronic device.

The computer readable medium carries one or more programs which, when executed by the electronic device, cause the electronic device to: acquiring search associated information corresponding to the search terms; determining a plurality of associated words of the search word according to the search associated information; determining a target value parameter of each associated word, wherein the target value parameter is used for representing value information corresponding to the clicked associated word and value information corresponding to the displayed associated word; and determining a recommended search word from the associated words according to the target value parameter of the associated word.

Computer program code for carrying out operations for the present disclosure may be written in any combination of one or more programming languages, including but not limited to an object oriented programming language such as Java, Smalltalk, C + +, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider).

The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

The modules described in the embodiments of the present disclosure may be implemented by software or hardware. The name of the module does not in some cases form a limitation on the module itself, and for example, the obtaining module may also be described as a module for obtaining search related information corresponding to a search term.

The functions described herein above may be performed, at least in part, by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field Programmable Gate Arrays (FPGAs), Application Specific Integrated Circuits (ASICs), Application Specific Standard Products (ASSPs), systems on a chip (SOCs), Complex Programmable Logic Devices (CPLDs), and the like.

In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

According to one or more embodiments of the present disclosure, there is provided a recommended search term determination method including:

acquiring search associated information corresponding to the search terms;

According to one or more embodiments of the present disclosure, there is provided a recommended search word determining method, the search related information including a plurality of items of related information;

the determining a plurality of associated words of the search word according to the search associated information includes:

determining the relevance between each piece of relevant information and the search word based on a relevance model;

determining target association information of which the association degree is greater than an association degree threshold value in the plurality of items of association information;

performing word segmentation processing on the target associated information, and determining the importance degree corresponding to each word, wherein the importance degree is the product of the word frequency of the word appearing in the target associated information and the inverse document frequency of the word in the corpus;

determining a core word of the target associated information according to the corresponding importance of each word segmentation;

and acquiring a plurality of associated words of the search word based on the core word.

According to one or more embodiments of the present disclosure, there is provided a recommended search word determining method, where the obtaining of multiple associated words of the search word based on the core word includes:

According to one or more embodiments of the present disclosure, there is provided a recommended search term determining method for determining a target value parameter of the related term by:

acquiring click value information of the associated word;

determining a first value parameter of the associated word according to the click value information, wherein the first value parameter is used for representing value information corresponding to the clicked associated word;

acquiring display value information of the associated word, interactive data of the associated word in a target historical time period and a delivery return rate of the associated word in the target historical time period;

determining a second value parameter of the associated word according to the display value information, the interaction data and the delivery return rate, wherein the second value parameter is used for representing value information corresponding to the displayed associated word;

and determining the target value parameter of the associated word according to the first value parameter and the second value parameter.

According to one or more embodiments of the present disclosure, there is provided a recommended search term determination method, where the presentation value information is a single presentation value;

the determining a second value parameter of the associated word according to the display value information, the interaction data and the delivery return rate comprises:

According to one or more embodiments of the present disclosure, there is provided a method for determining a recommended search term, where determining a recommended search term from a plurality of associated terms according to a target value parameter of the associated term includes:

and determining the associated word corresponding to the maximum first N recommended values as a recommended search word according to the recommended value corresponding to the associated word, wherein N is a positive integer.

According to one or more embodiments of the present disclosure, there is provided a recommended search word determining method in which an interaction probability of a related word is determined by:

According to one or more embodiments of the present disclosure, there is provided a recommended search word determining method in which a related word having a correlation with the related word is obtained by:

According to one or more embodiments of the present disclosure, there is provided a recommended search term determination method, where the search related information includes at least one of: the search result information corresponding to the search word, the interaction behavior information associated with the search word, and the search behavior information associated with the search word.

According to one or more embodiments of the present disclosure, there is provided a recommended search term determination apparatus including:

According to one or more embodiments of the present disclosure, there is provided a computer-readable medium on which a computer program is stored, which when executed by a processing device, implements the steps of the recommended search term determination method according to any embodiment of the present disclosure.

According to one or more embodiments of the present disclosure, there is provided an electronic device including:

a storage device having a computer program stored thereon;

processing means for executing the computer program in the storage means to implement the steps of the recommended search term determination method according to any embodiment of the present disclosure.

The foregoing description is only exemplary of the preferred embodiments of the disclosure and is illustrative of the principles of the technology employed. It will be appreciated by those skilled in the art that the scope of the disclosure herein is not limited to the particular combination of features described above, but also encompasses other embodiments in which any combination of the features described above or their equivalents does not depart from the spirit of the disclosure. For example, the above features and (but not limited to) the features disclosed in this disclosure having similar functions are replaced with each other to form the technical solution.

Further, while operations are depicted in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Likewise, while several specific implementation details are included in the above discussion, these should not be construed as limitations on the scope of the disclosure. Certain features that are described in the context of separate embodiments can also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment can also be implemented in multiple embodiments separately or in any suitable subcombination.

Although the subject matter has been described in language specific to structural features and/or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as example forms of implementing the claims. With regard to the apparatus in the above-described embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment related to the method, and will not be elaborated here.

Claims

1. A method for determining recommended search terms, the method comprising:

acquiring search associated information corresponding to the search terms;

2. The method of claim 1, wherein the search related information comprises a plurality of items of related information;

3. The method of claim 2, wherein the obtaining a plurality of associated words of the search word based on the core word comprises:

4. The method according to claim 1, wherein the target value parameter of the relevant word is determined by:

acquiring click value information of the associated word;

5. The method of claim 4, wherein the presentation value information is a single presentation value;

6. The method according to claim 1, wherein the determining a recommended search term from the plurality of related terms according to the target value parameter of the related term includes:

7. The method according to claim 6, wherein the interaction probability of the associated word is determined by:

8. The method according to claim 7, wherein the related word having a correlation with the related word is obtained by:

9. The method according to any one of claims 1-8, wherein the search related information comprises at least one of: the search result information corresponding to the search word, the interaction behavior information associated with the search word, and the search behavior information associated with the search word.

10. An apparatus for determining recommended search terms, the apparatus comprising:

11. A computer-readable medium, on which a computer program is stored, characterized in that the program, when being executed by processing means, carries out the steps of the method of any one of claims 1-9.

12. An electronic device, comprising:

a storage device having a computer program stored thereon;

processing means for executing the computer program in the storage means to carry out the steps of the method according to any one of claims 1 to 9.