Search intent determination method, device, equipment, and storage medium

Abstract

The present disclosure provides a method, device, equipment and storage medium for determining search intent, which relates to the field of artificial intelligence technology, in order to at least solve the problem of low accuracy of search intent determined in the prior art. The method comprises: obtaining a search session sequence, the search session sequence comprising at least two consecutive search sessions; for each search session in the search session sequence, obtaining the dimensional features corresponding to the search session, the dimensional features comprising the target features; the target features are obtained by estimating the search intent of the search session based on the pause time of the historical search sessions of the account corresponding to the search session; determining the intent label of the search session according to the dimensional features corresponding to the search session; the intent label is used to characterize the continuity between the search intent of the search session and the search intent of the adjacent search session.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, and in particular to a method, apparatus, device and storage medium for determining search intent. Background Art

[0002] Generally, when a user needs to know certain information, he or she will search on the Internet to obtain the corresponding information, wherein the corresponding information is determined according to the user's search intention.

[0003] In the related art, the user's search intent is mostly determined by the user's query time or query words. This determination method uses relatively simple features, making the final search intent less accurate. Summary of the invention

[0004] The present disclosure provides a method, apparatus, device and storage medium for determining search intent, so as to at least solve the problem of low accuracy of search intent determined in the prior art.

[0005] The technical solution of the present disclosure is as follows:

[0006] According to a first aspect of the present disclosure, a method for determining search intent is provided, the method comprising: a search intent determination device obtains a search session sequence, the search session sequence comprising at least two consecutive search sessions; for each search session in the search session sequence, a dimensional feature corresponding to the search session is obtained, the dimensional feature comprising a target feature; the target feature is obtained by estimating the search intent of the search session based on the pause time of the historical search sessions of the account corresponding to the search session; based on the dimensional feature corresponding to the search session, an intent label of the search session is determined; the intent label is used to characterize the continuity between the search intent of the search session and the search intent of an adjacent search session.

[0007] Optionally, the dimensional features corresponding to the search session also include at least one of the following: a common word length feature, a coincidence feature after word segmentation, and a similarity feature; the common word length feature includes: a first length parameter or a second length parameter, the first length parameter is used to characterize the ratio of the length value of the common word to the length value of the word of the adjacent search session, and the second length parameter is used to characterize the ratio of the length value of the common word to the length value of the word of the search session; the common word is a word shared by the search session and the adjacent search session; the coincidence feature after word segmentation includes: a first coincidence parameter or a second coincidence parameter, the first coincidence parameter is used to characterize the degree of coincidence of the common morpheme with the corresponding word segmentation of the adjacent search session, and the second coincidence parameter is used to characterize the degree of coincidence of the common morpheme with the word segmentation of the search session; the common morpheme is a word with the same attributes in the search session and the adjacent search session; the similarity feature includes: the text cosine distance between the search session and the adjacent search session.

[0008] Optionally, in the case where the adjacent search session includes a previous search session of the search session, the dimensional feature includes a first time interval, which is the time interval between the search session and the previous search session; in the case where the adjacent search session includes a subsequent search session of the search session, the dimensional feature includes a second time interval, which is the time interval between the search session and the subsequent search session.

[0009] Optionally, the intent label of each search session is determined based on the dimensional features corresponding to each search session, including: inputting the dimensional features corresponding to each search session into a search intent determination model to obtain the intent label of each search session, and the search intent determination model is a gradient boosting decision tree model.

[0010] Optionally, the method further includes: obtaining a search session sequence sample and a standard intent label for each search session sample in the search session sequence sample; wherein the standard intent label is a first label or a second label, the first label is used to characterize that the search session sample is the first search session sample in the search session sequence sample, and the first search session sample is the starting search session sample corresponding to the search intent, or the search intent of the search session sample is inconsistent with the search intent of the previous search session sample; the second label is used to characterize that the search intent of the search session sample is similar to the search intent of the previous search session sample; similar search intent includes the same search intent, a narrowed scope of the search intent, an expanded scope of the search intent, or a related search intent. A preset model is trained based on the search session sequence samples and the standard intent label to obtain a search intent determination model.

[0011] Optionally, the method also includes: based on the search session sequence and the intent tag corresponding to each search session, classifying the search sessions belonging to the same intent tag to obtain a search session library; or, based on the search session sequence and the intent tag corresponding to each search session, integrating the search results corresponding to the search sessions belonging to the same intent tag to obtain a search result library.

[0012] Optionally, the method also includes: obtaining a first search session and a second search session, both of which are continuous search sessions with the same intent tag in a search session sequence; obtaining a first number of search results corresponding to the first search session, and a second number of search results corresponding to the second search session; and determining an intent relationship between the first search session and the second search session based on a relationship between the first number of entries and the second number of entries.

[0013] Optionally, the method also includes: when the number of first entries is greater than the number of second entries, determining that the scope of the search intent of the first search session is greater than the scope of the search intent of the second search session; when the number of first entries is less than the number of second entries, determining that the scope of the search intent of the first search session is smaller than the scope of the search intent of the second search session.

[0014] Optionally, the method further includes: obtaining the number of consecutive search sessions with the same intent tag in the search session sequence; and determining the user satisfaction based on the number.

[0015] According to the second aspect of the present disclosure, a search intent determination device is provided, which includes an acquisition module and a processing module. The acquisition module is configured to acquire a search session sequence, which includes at least two consecutive search sessions; the processing module is configured to acquire, for each search session in the search session sequence, dimensional features corresponding to the search session, which dimensional features include target features; the target feature is obtained by estimating the search intent of the search session based on the pause time of the historical search sessions of the account corresponding to the search session; the processing module is also configured to determine the intent label of the search session based on the dimensional features corresponding to the search session; the intent label is used to characterize the continuity between the search intent of the search session and the search intent of the adjacent search session.

[0016] Optionally, the dimensional features corresponding to the search session also include at least one of the following: a common word length feature, a coincidence feature after word segmentation, and a similarity feature; the common word length feature includes: a first length parameter or a second length parameter, the first length parameter is used to characterize the ratio of the length value of the common word to the length value of the word of the adjacent search session, and the second length parameter is used to characterize the ratio of the length value of the common word to the length value of the word of the search session; the common word is a word shared by the search session and the adjacent search session; the coincidence feature after word segmentation includes: a first coincidence parameter or a second coincidence parameter, the first coincidence parameter is used to characterize the degree of coincidence of the common morpheme with the corresponding word segmentation of the adjacent search session, and the second coincidence parameter is used to characterize the degree of coincidence of the common morpheme with the word segmentation of the search session; the common morpheme is a word with the same attributes in the search session and the adjacent search session; the similarity feature includes: the text cosine distance between the search session and the adjacent search session.

[0017] Optionally, in the case where the adjacent search session includes a previous search session of the search session, the dimensional feature includes a first time interval, which is the time interval between the search session and the previous search session; in the case where the adjacent search session includes a subsequent search session of the search session, the dimensional feature includes a second time interval, which is the time interval between the search session and the subsequent search session.

[0018] Optionally, the processing module is also configured to input the dimensional features corresponding to each search session into the search intent determination model to obtain the intent label of each search session, and the search intent determination model is a gradient boosting decision tree model.

[0019] Optionally, the acquisition module is further configured to acquire the search session sequence sample and the standard intent label of each search session sample in the search session sequence sample; wherein the standard intent label is a first label or a second label, the first label is used to characterize that the search session sample is the first search session sample in the search session sequence sample, and the first search session sample is the starting search session sample corresponding to the search intent, or the search intent of the search session sample is inconsistent with the search intent of the previous search session sample; the second label is used to characterize that the search intent of the search session sample is similar to the search intent of the previous search session sample; similar search intent includes the same search intent, a narrowed scope of search intent, an expanded scope of search intent, or related search intent. The processing module is further configured to train a preset model based on the search session sequence sample and the standard intent label to obtain a search intent determination model.

[0020] Optionally, the processing module is further configured to classify search sessions with the same intent tag based on the search session sequence and the intent tag corresponding to each search session to obtain a search session library; or, based on the search session sequence and the intent tag corresponding to each search session, integrate the search results corresponding to the search sessions with the same intent tag to obtain a search result library. The acquisition module is further configured to acquire a first search session and a second search session, both of which are continuous search sessions with the same intent tag in the search session sequence; the acquisition module is further configured to acquire a first number of search results corresponding to the first search session and a second number of search results corresponding to the second search session; the processing module is further configured to determine the intent relationship between the first search session and the second search session based on the relationship between the first number of entries and the second number of entries.

[0021] Optionally, the processing module is further configured to determine that the scope of the search intent of the first search session is larger than the scope of the search intent of the second search session when the number of first entries is larger than the number of second entries; the processing module is further configured to determine that the scope of the search intent of the first search session is smaller than the scope of the search intent of the second search session when the number of first entries is smaller than the number of second entries.

[0022] Optionally, the acquisition module is further configured to obtain the number of consecutive search sessions with the same intent tag in the search session sequence; the processing module is further configured to determine the user's satisfaction based on the number.

[0023] According to a third aspect of the present disclosure, an electronic device is provided, comprising: a processor and a memory for storing instructions executable by the processor; wherein the processor is configured to execute instructions to implement any optional search intent determination method as described in the first aspect above.

[0024] According to a fourth aspect of the present disclosure, a computer-readable storage medium is provided, on which instructions are stored. When the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute any optional search intent determination method as described in the first aspect above.

[0025] According to a fifth aspect of the present disclosure, a computer program product is provided, comprising instructions. When the instructions in the computer program product are executed by a processor of an electronic device, any optional search intent determination method as described in the first aspect above is implemented.

[0026] The technical solution provided by the embodiments of the present disclosure brings at least the following beneficial effects:

[0027] In the above scheme, the electronic device obtains a search session sequence and determines the dimensional features of each search session in the search session sequence. The dimensional features include target features. The target features are obtained by estimating the search intention based on the time features of the historical search sessions of the account corresponding to the search session. Therefore, the intention label of the search session determined based on the target features will be more accurate.

[0028] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The drawings herein are incorporated into and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the description are used to explain the principles of the present disclosure, and do not constitute improper limitations on the present disclosure.

[0030] Figure 1 is one of the flow charts of a method for determining search intent according to an exemplary embodiment;

[0031] Figure 2 is a second flowchart of a method for determining search intent according to an exemplary embodiment;

[0032] Figure 3 is a third flowchart of a method for determining search intent according to an exemplary embodiment;

[0033] Figure 4 is a fourth flowchart of a method for determining search intent according to an exemplary embodiment;

[0034] Figure 5 is a fifth flowchart of a method for determining search intent according to an exemplary embodiment;

[0035] Figure 6 is a sixth flowchart of a method for determining search intent according to an exemplary embodiment;

[0036] Figure 7 is a structural block diagram of a search intent determination device according to an exemplary embodiment;

[0037] Figure 8 It is a structural diagram of a computer program product of a search intent determination method provided according to an exemplary embodiment. DETAILED DESCRIPTION

[0038] In order to enable ordinary persons in the art to better understand the technical solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings.

[0039] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The data involved in the present disclosure may be data authorized by the user or fully authorized by all parties. The implementation methods described in the following exemplary embodiments do not represent all implementation methods consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the attached claims.

[0040] It should also be understood that the term “comprising” indicates the presence of described features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements and / or components.

[0041] Based on the background technology, the embodiment of the present disclosure provides a method for determining search intent. The method is that an electronic device obtains a search session sequence and determines the dimensional features of each search session in the search session sequence, wherein the dimensional features include a target feature, and the target feature is obtained by estimating the search intent based on the time features of the historical search sessions of the account corresponding to the search session, so the intent label of the search session determined based on the target feature will be more accurate.

[0042] The following is an exemplary description of the search intent determination method provided in the embodiment of the present disclosure:

[0043] The search intent determination method provided by the present disclosure can be applied to electronic devices.

[0044] In some embodiments, the electronic device may be a server, a terminal, or other electronic device used for model training, which is not limited in the present disclosure.

[0045] The server may be a single server, or a server cluster consisting of multiple servers. In some implementations, the server cluster may also be a distributed cluster. The present disclosure does not limit the specific implementation of the server.

[0046] The terminal may be a mobile phone, tablet computer, desktop, laptop, handheld computer, notebook computer, ultra-mobile personal computer (UMPC), netbook, cellular phone, personal digital assistant (PDA), augmented reality (AR) and virtual reality (VR) devices, etc., which can install and use content community applications (such as Kuaishou). The present disclosure does not impose any special restrictions on the specific form of the electronic device. It can interact with the user through one or more methods such as keyboard, touchpad, touch screen, remote control, voice interaction or handwriting device.

[0047] The following will be combined with the drawings in the embodiments of the present disclosure to clearly and completely describe the technical solutions in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present disclosure.

[0048] like Figure 1 As shown, when the search intention determination method is applied to an electronic device, the search intention determination method may include:

[0049] S11. The electronic device obtains a search session sequence.

[0050] The search session sequence includes at least two consecutive search sessions.

[0051] In one embodiment, the search session sequence acquired by the electronic device is: multiple continuous search sessions input by the user within a period of time. The search session includes information such as a user account, a timestamp, and a search term. Exemplarily, the user account can be a user ID, which is used to classify search sessions belonging to the same user. The timestamp is the input time of the search session, which is used to provide a time feature for the judgment of the search intent. The search term is used to characterize the recall resources that you want to obtain. It should be noted that the multiple search sessions include search session A and search session B. Continuous means that the user enters search session B before or after entering search session A, and there may be a pause time between search session A and search session B, or there may be no pause time, but it is not limited to there must be no pause time between search session A and search session B.

[0052] S12. The electronic device obtains, for each search session in the search session sequence, a dimensional feature corresponding to the search session.

[0053] The dimensional features include target features; the target features are obtained by estimating the search intent of the search session based on the pause time of the historical search sessions of the account corresponding to the search session.

[0054] In one embodiment, the electronic device extracts dimensional features of each search session in the search session sequence. Specifically, the target features in the dimensional features are obtained by the electronic device inputting the search sessions in the search session sequence into a kernel density estimation model (KDE). KDE uses a non-parametric estimation method of probability function, which can form a probability function based on different search habits of different users, and then derive the target features of the search session based on the probability function.

[0055] S13. The electronic device determines an intention label of the search session according to the dimensional features corresponding to the search session, wherein the intention label is used to characterize the continuity between the search intention of the search session and the search intention of an adjacent search session.

[0056] In one embodiment, after obtaining the dimensional features corresponding to the search session, the electronic device normalizes the dimensional features to obtain the search intent label of each search session. Exemplarily, the intent label can be marked with B or I, where B is used to indicate that the search intent of a search session is inconsistent with the search intent of the previous search session, that is, the beginning of a new search intent, and I is used to indicate the middle of a search intent, which means that the search intent corresponding to the search session is inconsistent with the search intent of the previous search session. Figure 1 To.

[0057] The technical solution provided by the above embodiment has at least the following beneficial effects: by adopting the above technical features, the dimensional features of each search session in the search session sequence can be obtained, and the dimensional features represent the detailed information describing the search session. Then, based on the dimensional features corresponding to the search session, the intent label of the search session obtained will be more accurate.

[0058] Optionally, the dimensional features corresponding to the search session also include at least one of the following: a common word length feature, a coincidence feature after word segmentation, and a similarity feature.

[0059] The common word length feature includes: a first length parameter or a second length parameter, the first length parameter is used to characterize the ratio of the length value of the common word to the length value of the word of the adjacent search session, and the second length parameter is used to characterize the ratio of the length value of the common word to the length value of the word of the search session; the common word is a word shared by the search session and the adjacent search session;

[0060] The overlap feature after word segmentation includes: a first overlap parameter or a second overlap parameter, the first overlap parameter is used to characterize the overlap degree of the common morpheme relative to the word segmentation corresponding to the adjacent search session, and the second overlap parameter is used to characterize the overlap degree of the common morpheme relative to the word segmentation of the search session; the common morpheme is a word with the same attribute in the search session and the adjacent search session;

[0061] The similarity features include: the text cosine distance between the search session and the adjacent search sessions.

[0062] In some embodiments, the adjacent search session is a previous search session or a subsequent search session of the search session. For a search session i in the search session sequence, the similarity feature of the search session i includes: a text cosine distance between the search session i and the search session i+1 or a text cosine distance between the search session i and the search session i-1.

[0063] The common word length feature of search session i includes: a first length parameter or a second length parameter, wherein the first length parameter is obtained by dividing the length of the common words in search session i and search session i-1 (or search session i+1) by the length of the corresponding words in search session i-1 (or search session i+1). The second length parameter is specifically obtained by dividing the length of the common words in search session i and search session i-1 (or search session i+1) by the length of the corresponding words in search session i.

[0064] Remove irrelevant symbols such as commas and periods from search session i and search session i-1 (or search session i+1), and then segment search session i and search session i-1 (or search session i+1) according to the segmentation rule to obtain the segmentation result. The segmentation rule refers to placing words with the same attributes together. After obtaining the segmentation result, the overlap feature of search session i after segmentation includes: a first overlap parameter or a second overlap parameter. The first overlap is obtained by dividing the number of common morphemes in search session i and search session i-1 (or search session i+1) by the segmentation corresponding to search session i-1 (or search session i+1). The second overlap is obtained by dividing the number of common morphemes in search session i and search session i-1 (or search session i+1) by the segmentation corresponding to search session i.

[0065] For example, for search session i "I got admitted to Peking University", search session i-1 "I got admitted to Tsinghua University" and search session i+1 "Thursday", the similarity characteristics of search session i are: processing "I got admitted to Peking University", "I got admitted to Tsinghua University" and "Thursday" into word vector form, and calculating the cosine distance between the word vector corresponding to "I got admitted to Peking University" and the word vector corresponding to "Thursday", and calculating the cosine distance between the word vector corresponding to "I got admitted to Peking University" and the word vector corresponding to "I got admitted to Tsinghua University".

[0066] The length of the common words between search session i "I was admitted to Peking University" and search session i-1 "I was admitted to Tsinghua University" is 4. For the common word length feature of search session i, the first length parameter is the length of the same word in search session i and search session i-1 divided by the length of the word in search session i-1 = 4 / 6 = 2 / 3; the second length parameter is the length of the same word in search session i and search session i-1 divided by the length of the word in search session i = 4 / 6 = 2 / 3.

[0067] "I was admitted to Peking University" is divided into three words according to its attributes, namely, the subject "I", the predicate "kaoshang", the object "Peking University" and the object complement "le". "I was admitted to Tsinghua University" is divided into three words, namely, the subject "I", the predicate "kaoshang", the object "Tsinghua" and the object complement "le". For the overlap feature after word segmentation of search session i, the first overlap parameter is the number of common morphemes in search session i and search session i-1 divided by the number of search session i-1 after word segmentation = 3 / 4, and the second overlap parameter is the number of common morphemes in search session i and search session i-1 divided by the number of search session i after word segmentation = 3 / 4.

[0068] The technical solution provided by the above embodiment has at least the following beneficial effects: the dimensional features add three options: common word length feature, overlap feature after word segmentation and similarity feature. By adding these three features, the search session can be further analyzed in depth to facilitate subsequent acquisition of more accurate intent tag results.

[0069] Optionally, in a case where the adjacent search session includes a previous search session of the search session, the dimensional feature includes a first time interval, and the first time interval is a time interval between the search session and the previous search session.

[0070] In the case where the adjacent search session includes a subsequent search session of the search session, the dimension feature includes a second time interval, and the second time interval is a time interval between the search session and the subsequent search session.

[0071] In some embodiments, after the occurrence time of search session i, the occurrence time of search session i+1, and the occurrence time of search session i-1 in the search session sequence are clarified, a first time interval can be determined based on the occurrence time of search session i and the occurrence time of search session i-1, and the first time interval is the pause time between search session i and search session i-1. A second time interval can be determined based on the occurrence time of search session i and the occurrence time of search session i+1, and the second time interval is the pause time between search session i and search session i+1. The difference between the first time interval and the second time interval can also be used as a reference factor for evaluating whether they belong to the same search intention.

[0072] The technical solution provided by the above embodiment has at least the following beneficial effects: by adopting the above technical features, the dimensional features also include a first time interval and a second time interval, and the difference between the first time interval and the second time interval can also be used to determine the search intent of the search session, so as to further accurately determine the intention label of the search session.

[0073] For example, Figure 2 As shown, S13 determines the intent label of each search session according to the dimensional features corresponding to each search session, and also includes:

[0074] S131. The electronic device inputs the dimensional features corresponding to each search session into a search intent determination model to obtain an intent label for each search session.

[0075] Among them, the search intent determination model is the Gradient Boosting Decision Tree model (Gradient Boosting Decision Tree, GBDT).

[0076] In one embodiment, after the electronic device obtains the corresponding dimensional features of the search session, it inputs them into the search intent determination model, and the search intent determination model normalizes the difference between the first time interval and the second time interval in the dimensional features, normalizes the common word length features of the search session, and normalizes the coincidence features after the search session segmentation to obtain the processing results. The processing results are then combined with the target features in the dimensional features and the similarity features of the search session to output the intent label of the search session. The obtained intent label can be used to characterize user satisfaction scenarios, optimize search result page scenarios, cluster different search sessions belonging to the same search intent, and correct search sessions with writing errors, etc., and the present disclosure does not limit this.

[0077] Exemplarily, the search session sequence acquired by the electronic device includes multiple search sessions such as "Ultraman cartoon", "Zhang San", "Zhang San fishing", "Zhang San fishing monitoring", "Li Si", "Rich people's life", "Rich people's life", "Tyrant life", "Mobile screenshot" and "How to take screenshot on Apple mobile phone", etc. These search sessions are input into the search intention determination model. The search intention determination model can determine through judgment that "Ultraman cartoon" belongs to a new search intention; "Zhang San" belongs to a new search intention; "Zhang San fishing" and "Zhang San fishing monitoring" belong to the same intention and are related to the search intention corresponding to "Zhang San"; "Li Si" belongs to a new search intention; "Rich people's life" belongs to a new search intention; "Rich people's life" and "Tyrant life" belong to the same intention and are related to the search intention corresponding to "Rich people's life"; "Mobile screenshot" belongs to a new search intention; "How to take screenshot on Apple mobile phone" and "Mobile screenshot" belong to the same intention. Finally, the search intention determination model outputs the intention label results shown in Table 1.

[0078] Table 1

[0079]

[0080] The technical solution provided by the above embodiment has at least the following beneficial effects: by adopting the above technical features, multiple features in the dimensional features are processed through the search intent determination model, the influence of abnormal data on the results is avoided, and an intent label with higher accuracy is obtained.

[0081] Optionally, after the search intent label of each search session is determined, the effectiveness of the search intent determination model may be judged based on the accuracy and recall of the search intent label.

[0082] The accuracy rate indicates how many samples predicted as new search intents are truly new search intents. There are two possibilities for predicting new search intents: one is to predict new search intents as new search intents (TP), and the other is to predict non-new search intents as new search intents (FP). Then, the accuracy rate satisfies the expression: Accuracy = TP / (TP+FP).

[0083] The recall rate indicates how many new search intents are predicted correctly for a search session. There are two situations in which a new search intent is predicted correctly. One is to predict the original new search intent as a new search intent (TP), and the other is to predict the original new search intent as a non-new search intent (FN). Then, the recall rate satisfies the following expression: recall rate = TP / (TP+FN). Some experiments show that using the GBDT method, the accuracy of the intent label of the search session sequence is 0.9765, the recall rate is 0.9735, and the result of the accuracy and recall fitting is 0.9748. These results all show that the use of the GBDT method has a significant effect.

[0084] In one embodiment, in combination Figure 1 ,like Figure 3 As shown, the method also includes:

[0085] S31. The electronic device obtains a search session sequence sample and a standard intent label for each search session sample in the search session sequence sample.

[0086] Among them, the standard intent label is the first label or the second label, the first label is used to characterize that the search session sample is the first search session sample in the search session sequence sample, and the first search session sample is the starting search session sample corresponding to the search intent, or the search intent of the search session sample is inconsistent with the search intent of the previous search session sample; the second label is used to characterize that the search intent of the search session sample is similar to the search intent of the previous search session sample; similar search intent includes the same search intent, a narrowed scope of search intent, an expanded scope of search intent, or related search intent.

[0087] In one embodiment, the search session sequence samples are all search session samples searched by a user within a period of time, and the standard intent label of each search session sample is the result of labeling each search session sample.

[0088] Specifically, the standard intent label of each search session sample is determined in the following manner: According to the session time and search terms of the search session sequence sample, the search session sequence sample is cut to obtain multiple sub-search sessions; by judging whether the search intent of each sub-search session is consistent with the search intent of the previous sub-search intent, the intent label of each sub-search session is obtained.

[0089] Exemplarily, the first tag in the standard intent tag is used to indicate that the two search sessions are completely unrelated. For example, the first search session is: King of Glory, and the second search session is: Zhang San. The second tag in the standard intent tag is used to indicate the search intent type of the two search sessions. Similar search intents include repeated search intent, refined search intent, generalized search intent, related search intent, or reconstructed search intent. Among them, repeated search intent is used to indicate that two consecutive search sessions are completely consistent; for example, the first search session is: King of Glory, and the second search session is also: King of Glory. Refined search intent is used to indicate that the query term of the second search session is more detailed than that of the first search session; for example, the first search session is: King of Glory, and the second search session is: King of Glory Liu Bei Jungler Attached Text. Generalized search intent is used to indicate that the query term of the second search session is more rough and has a larger range than that of the first search session; for example, the first search session is: King of Glory Liu Bei Jungler Attached Text, and the second search session is: King of Glory. Related search intent is used to indicate that the keywords in the two search sessions include some of the same words; for example, the first search session is: IQ test, and the second search session is: intelligence test. Reconstructed search intent is used to indicate that the meanings expressed in two previous and subsequent search sessions are similar; for example: the first search session is: the capital of China, and the second search session is: Beijing.

[0090] S32. The electronic device trains a preset model based on the search conversation sequence samples and standard intent labels to obtain a search intent determination model.

[0091] Specifically, the electronic device inputs each search session sample into a preset model, and the preset model outputs a sample intent label of each search session sample, and then compares the sample intent label of each search session sample with the corresponding standard intent label to determine the difference between the two; adjusts the parameters of the preset model according to the difference value, and inputs the search session sample again after adjustment, and repeats the above operation until the difference value obtained is less than or equal to the preset threshold. At this time, the parameters in the preset model are considered to be standard parameters, and the preset model is also considered to be a search intent determination model.

[0092] The technical solution provided by the above embodiment has at least the following beneficial effects: by adopting the above technical features, a large amount of sample data is used to train the preset model, so as to obtain a search intent determination model that meets the requirements.

[0093] In one embodiment, if Figure 4 As shown, the method also includes:

[0094] S41a: The electronic device classifies the search sessions belonging to the same intent tag based on the search session sequence and the intent tag corresponding to each search session to obtain a search session library.

[0095] S41b: The electronic device integrates the search results corresponding to the search sessions belonging to the same intent tag based on the search session sequence and the intent tag corresponding to each search session to obtain a search result library.

[0096] In one embodiment, a sequence of search sessions with annotated intent tags is obtained. Adjacent search sessions with the same intent tag are classified to obtain a search session library. The search session library includes multiple search session sub-libraries, and the intent tags of the search sessions in each search session sub-library are consistent. The search results corresponding to the search sessions with the same intent tag are also classified and sorted to obtain a search result library, which includes multiple search result sub-libraries, and the search results in each search result sub-library have similar intents. Each search session sub-library in the multiple search session sub-libraries establishes a corresponding relationship with each search result sub-library based on the search intent.

[0097] In practical applications, the reasonable use of the search session library, the search result library and the corresponding relationship between the search session library and the search result library can greatly improve the work efficiency of the staff.

[0098] In one embodiment, in combination Figure 1 ,like Figure 5 As shown, the method also includes:

[0099] S51: The electronic device obtains a first search session and a second search session.

[0100] The first search session and the second search session are both continuous search sessions with the same intent tag in the search session sequence.

[0101] Exemplarily, the first search session and the second search session may be two consecutive search sessions in a search session sub-library in the search session library, or may be consecutive search sessions with the same intent input by the current user acquired by the electronic device.

[0102] S52: The electronic device obtains a first number of search results corresponding to the first search session and a second number of search results corresponding to the second search session.

[0103] Exemplarily, the first number of entries may be obtained based on the search results corresponding to the first search session in the search results page, or may be obtained based on the corresponding relationship between the search session library and the search results library. The second number of entries may be obtained based on the search results corresponding to the second search session in the search results page, or may be obtained based on the corresponding relationship between the search session library and the search results library.

[0104] S53: The electronic device determines an intention relationship between the first search session and the second search session according to a relationship between the first number of entries and the second number of entries.

[0105] Exemplary, combined Figure 5 S53: The electronic device determines the intention relationship between the first search session and the second search session according to the relationship between the first number of entries and the second number of entries, including:

[0106] S531a: When the first number of entries is greater than the second number of entries, determine that the scope of the search intent of the first search session is greater than the scope of the search intent of the second search session.

[0107] S531b: When the first number of entries is smaller than the second number of entries, determine that the scope of the search intent of the first search session is smaller than the scope of the search intent of the second search session.

[0108] Since the first search session and the second search session are search sessions with the same intent tag, the search results of the first search session and the second search session must have overlapping parts. Based on the first number of entries and the second number of entries, it can be determined whether the search range corresponding to the first search session is larger or the search range of the second search session is larger. The larger the search range, the broader the intent corresponding to the search session, and the smaller the search range, the more specific the intent corresponding to the search session. Based on the size relationship of the search range, the intent correlation is established for the first search session and the second search session. After determining the correlation of the scope of the search intent, the first search session and the second search session can be further analyzed based on the correlation.

[0109] For example, the first search session is "Glory of the King", and the second search session is "Heroes of the King". The first number of entries corresponding to the first search session is 10,000. The second number of entries corresponding to the second search session is 5,000. Based on the number of entries, it is considered that the scope level of the search intent of the first search session is higher than the scope level of the search intent of the second search session.

[0110] Exemplary, combined Figure 6 , the method of the present disclosure further comprises:

[0111] S61. Obtain the number of consecutive search sessions with the same intent tag in a search session sequence.

[0112] S62. Determine user satisfaction based on the quantity.

[0113] In some embodiments, the number of consecutive search sessions with the same intent tag is counted, and the user satisfaction is determined based on the search session data.

[0114] Exemplarily, search session 1 is "Honor of Kings", search session 2 is "Heroes of Honor of Kings", search session 3 is "Hero Liu Bei of Honor of Kings", and search session 4 is "Honors of Kings Liu Bei Jungler Attachment". Search sessions 1 to 4 are all related to Honor of Kings and have the same intent tag. The user determines the final desired search result through 4 searches. The judgment rule of user satisfaction is set based on quantity. The judgment rule can be that when the number of consecutive search sessions with the same intent tag searched by the user is less than or equal to 2 times, the user satisfaction is considered to be 90%. When the number of user searches is greater than 2 times, the user satisfaction decreases by 5% for each additional search. It can also be to set a preset threshold. When the number of search sessions exceeds the preset threshold, the user is considered to be dissatisfied. When the number of search sessions does not exceed the preset threshold, the user is considered to be satisfied. The judgment rule can also be other rules, which are not limited by the present disclosure.

[0115] Combination of the above Figure 1-Figure 6 The method provided by the embodiment of the present disclosure is described in detail. In order to achieve the above functions, the search intention determination device includes hardware structures and / or software modules corresponding to the execution of each function, and these hardware structures and / or software modules corresponding to the execution of each function can constitute an electronic device. It should be easily appreciated by those skilled in the art that, in combination with the algorithm steps of each example described in the embodiments disclosed herein, the present disclosure can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present disclosure.

[0116] The embodiments of the present disclosure may divide the electronic device into functional modules according to the above method examples. For example, the electronic device may include a search intent determination device. The search intent determination device may divide each functional module corresponding to each function, or may integrate two or more functions into one processing module. The above integrated modules may be implemented in the form of hardware or in the form of software functional modules. It should be noted that the division of modules in the embodiments of the present disclosure is schematic and is only a logical functional division. There may be other division methods in actual implementation.

[0117] The following, combined Figure 7The search intent determination device provided by the embodiment of the present disclosure is described in detail. It should be understood that the description of the device embodiment corresponds to the description of the method embodiment, so the contents not described in detail can be referred to the method embodiment above, and for the sake of brevity, they will not be repeated here.

[0118] Figure 7 is a schematic diagram of a structure of a search intention determination device according to an exemplary embodiment, which is applied to an electronic device, see Figure 7 As shown, the search intention determination device includes: an acquisition module 71 and a processing module 72.

[0119] The acquisition module 71 is configured to acquire a search session sequence, where the search session sequence includes at least two consecutive search sessions; for example, referring to Figure 1 As shown, the acquisition module 71 is configured to execute S11.

[0120] The processing module 72 is configured to obtain, for each search session in the search session sequence, a dimensional feature corresponding to the search session, wherein the dimensional feature includes a target feature; the target feature is obtained by estimating the search intent of the search session based on the pause time of the historical search session of the account corresponding to the search session; for example, referring to Figure 1 As shown, the processing module 72 is configured to execute S12.

[0121] The processing module 72 is further configured to determine the intent label of the search session according to the dimensional features corresponding to the search session; the intent label is used to characterize the continuity between the search intent of the search session and the search intent of the adjacent search session. Figure 1 As shown, the processing module 72 is configured to execute S13.

[0122] Optionally, the dimensional features corresponding to the search session further include at least one of the following: a common word length feature, a coincidence feature after word segmentation, and a similarity feature;

[0123] The common word length feature includes: a first length parameter or a second length parameter, the first length parameter is used to characterize the ratio of the length value of the common word to the length value of the word of the adjacent search session, and the second length parameter is used to characterize the ratio of the length value of the common word to the length value of the word of the search session; the common word is a word shared by the search session and the adjacent search session;

[0124] The overlap feature after word segmentation includes: a first overlap parameter or a second overlap parameter, the first overlap parameter is used to characterize the overlap degree of the common morpheme relative to the word segmentation corresponding to the adjacent search session, and the second overlap parameter is used to characterize the overlap degree of the common morpheme relative to the word segmentation of the search session; the common morpheme is a word with the same attribute in the search session and the adjacent search session;

[0125] The similarity features include: the text cosine distance between the search session and the adjacent search sessions.

[0126] Optionally, in a case where the adjacent search session includes a previous search session of the search session, the dimension feature includes a first time interval, and the first time interval is a time interval between the search session and the previous search session;

[0127] In the case where the adjacent search session includes a subsequent search session of the search session, the dimension feature includes a second time interval, and the second time interval is a time interval between the search session and the subsequent search session.

[0128] Optionally, the processing module 72 is further configured to input the dimensional features corresponding to each search session into a search intent determination model to obtain an intent label for each search session, and the search intent determination model is a gradient boosting decision tree model. Figure 2 As shown, the processing module 72 is configured to execute S131.

[0129] Optionally, the acquisition module 71 is further configured to acquire the search session sequence sample and the standard intent label of each search session sample in the search session sequence sample; wherein the standard intent label is a first label or a second label, and the first label is used to characterize that the search session sample is the first search session sample in the search session sequence sample, and the first search session sample is the starting search session sample corresponding to the search intent, or the search intent of the search session sample is inconsistent with the search intent of the previous search session sample; the second label is used to characterize that the search intent of the search session sample is similar to the search intent of the previous search session sample; similar search intent includes the same search intent, a narrowed scope of the search intent, an expanded scope of the search intent, or a related search intent. For example, referring to Figure 3 As shown, the acquisition module 71 is configured to execute S31.

[0130] The processing module 72 is further configured to train a preset model based on the search session sequence samples and the standard intent labels to obtain a search intent determination model. Figure 3 As shown, the processing module 72 is configured to execute S32.

[0131] Optionally, the processing module 72 is further configured to classify the search sessions with the same intent tag based on the search session sequence and the intent tag corresponding to each search session to obtain a search session library; or, based on the search session sequence and the intent tag corresponding to each search session, integrate the search results corresponding to the search sessions with the same intent tag to obtain a search result library. Figure 4 As shown, the processing module 72 is configured to execute S41a and S41b.

[0132] Optionally, the acquisition module 71 is further configured to acquire a first search session and a second search session, where the first search session and the second search session are both continuous search sessions with the same intent tag in the search session sequence; for example, referring to Figure 5 As shown, the acquisition module 71 is configured to execute S51.

[0133] The acquisition module 71 is further configured to acquire a first number of search results corresponding to the first search session and a second number of search results corresponding to the second search session; for example, referring to Figure 5 As shown, the acquisition module 71 is configured to execute S52.

[0134] The processing module 72 is further configured to determine the intention relationship between the first search session and the second search session according to the relationship between the first number of entries and the second number of entries. Figure 5 As shown, the processing module 72 is configured to execute S53.

[0135] Optionally, the processing module 72 is further configured to determine that the scope of the search intent of the first search session is greater than the scope of the search intent of the second search session when the first number of entries is greater than the second number of entries; for example, referring to Figure 5 As shown, the processing module 72 is configured to execute S531a.

[0136] The processing module 72 is further configured to determine that the scope of the search intent of the first search session is smaller than the scope of the search intent of the second search session when the first number of entries is smaller than the second number of entries. Figure 5 As shown, the processing module 72 is configured to execute S531b.

[0137] Optionally, the acquisition module 71 is further configured to acquire the number of consecutive search sessions with the same intent tag in the search session sequence; for example, referring to Figure 6 As shown, the acquisition module 71 is configured to execute S61.

[0138] The processing module 72 is further configured to determine the user's satisfaction based on the quantity. Figure 1 As shown, the processing module 72 is configured to execute S62.

[0139] Of course, the search intent determination device provided by the embodiment of the present disclosure includes but is not limited to the above modules, for example, the search intent determination device may also include a storage module 73. The storage module 73 may be used to store the program code of the search intent determination device, and may also be used to store data generated by the search intent determination device during operation, such as data in a write request, etc.

[0140] In actual implementation, the specific execution process of the acquisition module 71 and the processing module 72 can be referred to Figure 1-Figure 6 The description of any of the search intent determination methods shown in the section will not be repeated here.

[0141] Another embodiment of the present disclosure further provides a computer-readable storage medium in which instructions are stored. When the instructions are executed on an electronic device, the electronic device executes the above-mentioned Figure 1-Figure 6 Any of the search intent determination methods shown in .

[0142] Figure 8 A conceptual partial view of a computer program product provided by an embodiment of the present disclosure is schematically shown, where the computer program product includes a computer program for executing a computer process on a computing device.

[0143] In one embodiment, the computer program product is provided using a signal bearing medium 810. The signal bearing medium 810 may include one or more program instructions that, when executed by one or more processors, may provide the above-described Figure 1 Thus, for example, reference to Figure 1 In the embodiment shown in , one or more features of S11 to S13 may be undertaken by one or more instructions associated with the signal bearing medium 810. In addition, Figure 8 The program instructions in also describe example instructions.

[0144] In some examples, the signal bearing medium 810 may include a computer readable medium 811 such as, but not limited to, a hard drive, a compact disk (CD), a digital video disk (DVD), a digital tape, a memory, a read-only memory (ROM) or a random access memory (RAM), and the like.

[0145] In some implementations, the signal bearing medium 810 may include a computer recordable medium 812 such as, but not limited to, a memory, a read / write (R / W) CD, a R / W DVD, or the like.

[0146] In some implementations, signal bearing medium 810 may include communication medium 813 such as, but not limited to, digital and / or analog communication media (eg, fiber optic cables, waveguides, wired communication links, wireless communication links, etc.).

[0147] The signal bearing medium 810 may be communicated by a wireless form of communication medium 813. The one or more program instructions may be, for example, computer executable instructions or logic implemented instructions.

[0148] In some examples, such as for Figure 7 The described search intent determination apparatus may be configured to provide various operations, functions, or actions in response to one or more program instructions via computer-readable media 811 , computer-recordable media 812 , and / or communication media 813 .

[0149] Through the description of the above implementation methods, technical personnel in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0150] In the several embodiments provided in the present disclosure, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of modules or units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0151] The units described as separate components may or may not be physically separated, and the components shown as units may be one physical unit or multiple physical units, that is, they may be located in one place or distributed in multiple different places. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0152] In addition, each functional unit in each embodiment of the present disclosure may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0153] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solution of the embodiment of the present disclosure is essentially or the part that contributes to the prior art or the full classification part or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a device (which can be a single-chip microcomputer, chip, etc.) or a processor (processor) to perform the full classification part or part of the steps of each embodiment method of the present disclosure. The aforementioned storage medium includes: U disk, mobile hard disk, ROM, RAM, disk or optical disk and other media that can store program code.

[0154] The above are only specific embodiments of the present disclosure, but the protection scope of the present disclosure is not limited thereto. Any changes or substitutions within the technical scope disclosed in the present disclosure should be included in the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be based on the protection scope of the claims.

Claims

1. A method for determining search intent, characterized in that: include: Acquire a search session sequence, wherein the search session sequence includes at least two consecutive search sessions input by a user; For each search session in the search session sequence, a dimensional feature corresponding to the search session is obtained, wherein the dimensional feature includes a target feature; the target feature is obtained by estimating the search intent of the search session based on the pause time of the historical search session of the account corresponding to the search session; the dimensional feature corresponding to the search session also includes at least one of the following: a common word length feature, a coincidence feature after word segmentation, and a similarity feature; Determining an intent tag of the search session according to the dimensional features corresponding to the search session; the intent tag is used to characterize the continuity between the search intent of the search session and the search intent of an adjacent search session; The common word length feature includes: a first length parameter or a second length parameter, the first length parameter is used to characterize the ratio of the length value of the common word to the length value of the word of the adjacent search session, and the second length parameter is used to characterize the ratio of the length value of the common word to the length value of the word of the search session; the common word is a word shared by the search session and the adjacent search session; The overlap feature after word segmentation includes: a first overlap parameter or a second overlap parameter, wherein the first overlap parameter is used to characterize the overlap degree of the common morpheme relative to the word segmentation corresponding to the adjacent search session, and the second overlap parameter is used to characterize the overlap degree of the common morpheme relative to the word segmentation of the search session; the common morpheme is a word with the same attribute in the search session and the adjacent search session; Acquire a first search session and a second search session, where the first search session and the second search session are both continuous search sessions with the same intent tag in the search session sequence; Obtaining a first number of search results corresponding to the first search session, and a second number of search results corresponding to the second search session; An intent relationship between the first search session and the second search session is determined based on a relationship between the first number of entries and the second number of entries.

2. The method according to claim 1, characterized in that The similarity feature includes: a text cosine distance between the search session and the adjacent search session.

3. The method according to claim 1, characterized in that In the case where the adjacent search session includes a previous search session of the search session, the dimensional feature includes a first time interval, and the first time interval is a time interval between the search session and the previous search session; In the case that the adjacent search session includes a subsequent search session of the search session, the dimensional feature includes a second time interval, where the second time interval is a time interval between the search session and the subsequent search session.

4. The method according to any one of claims 1 to 3, characterized in that The determining the intent label of each search session according to the dimensional features corresponding to each search session includes: The dimensional features corresponding to each search session are input into a search intent determination model to obtain an intent label for each search session, and the search intent determination model is a gradient boosting decision tree model.

5. The method according to claim 1, characterized in that The method further comprises: Obtain a search session sequence sample and a standard intent label for each search session sample in the search session sequence sample; wherein the standard intent label is a first label or a second label, the first label is used to characterize that the search session sample is the first search session sample in the search session sequence sample, and the first search session sample is the starting search session sample corresponding to the search intent, or the search intent of the search session sample is inconsistent with the search intent of the previous search session sample; the second label is used to characterize that the search intent of the search session sample is similar to the search intent of the previous search session sample; the similar search intent includes the same search intent, a narrowed scope of the search intent, an expanded scope of the search intent, or a related search intent; A preset model is trained according to the search session sequence samples and the standard intent labels to obtain the search intent determination model.

6. The method according to any one of claims 1 to 3, characterized in that The method further comprises: Based on the search session sequence and the intent tag corresponding to each search session, the search sessions belonging to the same intent tag are classified to obtain a search session library; or, Based on the search session sequence and the intent tag corresponding to each search session, search results corresponding to search sessions belonging to the same intent tag are integrated to obtain a search result library.

7. The method according to claim 1, characterized in that The determining, according to the relationship between the first number of entries and the second number of entries, the intention relationship between the first search session and the second search session includes: In a case where the first number of entries is greater than the second number of entries, determining that a scope of the search intent of the first search session is greater than a scope of the search intent of the second search session; In a case where the first number of entries is smaller than the second number of entries, it is determined that a scope of the search intent of the first search session is smaller than a scope of the search intent of the second search session.

8. The method according to any one of claims 1 to 3, characterized in that The method further comprises: Obtaining the number of consecutive search sessions with the same intent tag in the search session sequence; Based on the numbers, the user's satisfaction is determined.

9. A search intention determination device, characterized in that: include: An acquisition module is configured to acquire a search session sequence, wherein the search session sequence includes at least two consecutive search sessions input by a user; The processing module is configured to obtain, for each search session in the search session sequence, a dimensional feature corresponding to the search session, wherein the dimensional feature includes a target feature; the target feature is obtained by estimating the search intent of the search session based on the pause time of the historical search session of the account corresponding to the search session; The processing module is further configured to determine an intention label of the search session according to the dimensional features corresponding to the search session; the intention label is used to characterize the continuity between the search intention of the search session and the search intention of an adjacent search session; The dimensional features corresponding to the search session also include at least one of the following: a common word length feature, a coincidence feature after word segmentation, and a similarity feature; The common word length feature includes: a first length parameter or a second length parameter, the first length parameter is used to characterize the ratio of the length value of the common word to the length value of the word of the adjacent search session, and the second length parameter is used to characterize the ratio of the length value of the common word to the length value of the word of the search session; the common word is a word shared by the search session and the adjacent search session; The overlap feature after word segmentation includes: a first overlap parameter or a second overlap parameter, wherein the first overlap parameter is used to characterize the overlap degree of the common morpheme relative to the word segmentation corresponding to the adjacent search session, and the second overlap parameter is used to characterize the overlap degree of the common morpheme relative to the word segmentation of the search session; the common morpheme is a word with the same attribute in the search session and the adjacent search session; The acquisition module is further configured to acquire a first search session and a second search session, wherein the first search session and the second search session are both continuous search sessions with the same intent tag in the search session sequence; The acquisition module is further configured to acquire a first number of search results corresponding to the first search session, and a second number of search results corresponding to the second search session; The processing module is further configured to determine an intention relationship between the first search session and the second search session according to a relationship between the first number of entries and the second number of entries.

10. The device according to claim 9, characterized in that The similarity feature includes: a text cosine distance between the search session and the adjacent search session.

11. The device according to claim 10, characterized in that In the case where the adjacent search session includes a previous search session of the search session, the dimensional feature includes a first time interval, and the first time interval is a time interval between the search session and the previous search session; In the case that the adjacent search session includes a subsequent search session of the search session, the dimensional feature includes a second time interval, where the second time interval is a time interval between the search session and the subsequent search session.

12. The device according to any one of claims 9 to 11, characterized in that The processing module is also configured to input the dimensional features corresponding to each search session into a search intent determination model to obtain an intent label for each search session, and the search intent determination model is a gradient boosting decision tree model.

13. The device according to claim 9, characterized in that The acquisition module is further configured to acquire a search session sequence sample and a standard intent label of each search session sample in the search session sequence sample; wherein the standard intent label is a first label or a second label, the first label is used to characterize that the search session sample is the first search session sample in the search session sequence sample, and the first search session sample is the starting search session sample corresponding to the search intent, or the search intent of the search session sample is inconsistent with the search intent of the previous search session sample; the second label is used to characterize that the search intent of the search session sample is similar to the search intent of the previous search session sample; the similar search intent includes the same search intent, a narrowed scope of the search intent, an expanded scope of the search intent, or a related search intent; The processing module is also configured to train a preset model based on the search session sequence samples and the standard intent labels to obtain the search intent determination model.

14. The device according to any one of claims 9 to 11, characterized in that The processing module is further configured to classify search sessions belonging to the same intent tag based on the search session sequence and the intent tag corresponding to each search session to obtain a search session library; Or, based on the search session sequence and the intent tag corresponding to each search session, search results corresponding to search sessions belonging to the same intent tag are integrated to obtain a search result library.

15. The device according to claim 9, characterized in that The processing module is further configured to determine that a scope of the search intent of the first search session is greater than a scope of the search intent of the second search session when the first number of entries is greater than the second number of entries; The processing module is further configured to determine that a scope of the search intent of the first search session is smaller than a scope of the search intent of the second search session when the first number of entries is smaller than the second number of entries.

16. The device according to any one of claims 9 to 11, characterized in that The acquisition module is further configured to acquire the number of consecutive search sessions with the same intent tag in the search session sequence; The processing module is further configured to determine user satisfaction based on the quantity.

17. An electronic device, characterized in that: The electronic device comprises: processor; a memory for storing instructions executable by the processor; Wherein, the processor is configured to execute the instructions to implement the search intent determination method as described in any one of claims 1-8.

18. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the search intent determination method as described in any one of claims 1-8.

19. A computer program product comprising computer instructions, characterized in that When the computer instructions are executed by an electronic device, the search intent determination method as described in any one of claims 1-8 is implemented.

Images