A search intention recognition method, device, apparatus and storage medium
By dividing search requests into slots and labeling them by type, and using slot models and intent recognition models to identify search intent, the problem of insufficient precision and recall in intent recognition in complex search requests is solved, and more efficient intent recognition is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING ZITIAO NETWORK TECH CO LTD
- Filing Date
- 2023-09-19
- Publication Date
- 2026-06-30
AI Technical Summary
Existing technologies cannot effectively identify search intents that include complex search requests of various types, resulting in insufficient precision and recall in intent identification.
By dividing the search request into slots and labeling each slot segment with a type, the search intent of the search request is obtained by using a slot model and an intent recognition model to identify and fuse the intent of the slot segments respectively.
It improves the precision and recall of intent recognition for complex search requests, and solves the problem of insufficient precision and recall in intent recognition.
Smart Images
Figure CN117271601B_ABST
Abstract
Description
Technical Field
[0001] This disclosure relates to data processing technology, and more particularly to a search intent identification method, apparatus, device, and storage medium. Background Technology
[0002] After a user enters a search request, the search intent is determined by analyzing the search request. The relevance and accuracy of the search results depend on the accurate identification of the search intent.
[0003] For complex search requests containing multiple types of search terms, it is impossible to identify the search intent of the search request through cached queries. At the very least, the precision and recall of intent identification cannot meet the search requirements. Summary of the Invention
[0004] This disclosure provides a search intent recognition method, apparatus, device, and storage medium, which can improve the accuracy and recall rate of intent recognition for complex search requests.
[0005] In a first aspect, embodiments of this disclosure provide a method for identifying search intent, including:
[0006] The search request is divided into slots, and each slot segment is labeled with a type, wherein the slot represents the identifier of the target content in the search request that is associated with the search intent;
[0007] Intent recognition is performed on the slot segments based on the type of each slot segment;
[0008] The search intent of the search request is obtained by fusing the intent recognition results corresponding to each slot segment.
[0009] Secondly, embodiments of this disclosure also provide a search intent recognition device, the device comprising:
[0010] The slot partitioning module is used to partition the search request into slots and label each slot segment with a type. The slot represents the identifier of the target content in the search request that is associated with the search intent.
[0011] An intent recognition model is used to recognize the intent of each slot segment based on its type.
[0012] The result fusion module is used to fuse the intent recognition results corresponding to each slot segment to obtain the search intent of the search request.
[0013] Thirdly, embodiments of this disclosure also provide an electronic device, the electronic device comprising:
[0014] One or more processors;
[0015] Storage device for storing one or more programs.
[0016] When the one or more programs are executed by the one or more processors, the one or more processors implement the search intent recognition method as described in any embodiment of this disclosure.
[0017] Fourthly, embodiments of this disclosure also provide a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform the search intent recognition method as described in any embodiment of this disclosure.
[0018] This disclosure provides a search intent recognition method, apparatus, device, and storage medium. It involves dividing search requests into slots, labeling each slot segment with a type, recognizing intent based on the type of each slot segment, and then fusing the intent recognition results for each slot to obtain the search intent of the search request. By dividing search requests into slots and labeling them with types, this disclosure allows different types of slot segments to hit the cache separately, reducing the chance of complex search requests missing any cache entries. This increases the proportion of search requests with recognized search intent and improves the precision and recall rate of intent recognition for complex search requests, solving the problem that the precision and recall rate of intent recognition cannot meet search requirements. Attached Figure Description
[0019] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the originals and elements are not necessarily drawn to scale.
[0020] Figure 1 A flowchart illustrating a search intent recognition method provided in an embodiment of this disclosure;
[0021] Figure 2 A flowchart illustrating another search intent recognition method provided in this embodiment of the present disclosure;
[0022] Figure 3 This is a schematic diagram of the structure of a search intent recognition framework provided in an embodiment of the present disclosure;
[0023] Figure 4 This is a schematic diagram of the structure of a search intent recognition device provided in an embodiment of the present disclosure;
[0024] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. Detailed Implementation
[0025] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.
[0026] It should be understood that the steps described in the method embodiments of this disclosure may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of this disclosure is not limited in this respect.
[0027] The term "comprising" and its variations as used herein are open-ended inclusions, meaning "including but not limited to". The term "based on" means "at least partially based 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". Definitions of other terms will be given in the description below.
[0028] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.
[0029] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".
[0030] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.
[0031] It is understood that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and related provisions.
[0032] Figure 1 This is a flowchart illustrating a search intent recognition method provided in an embodiment of the present disclosure. This embodiment is applicable to situations where a complex search request is used to search for a search object. The method can be executed by a search intent recognition device, which can be implemented in software and / or hardware, or optionally, by an electronic device, such as a mobile terminal, a PC, or a server.
[0033] like Figure 1 As shown, the method includes:
[0034] S110. Divide the search request into slots and label the type of each slot segment.
[0035] The slot represents the identifier of the target content associated with the search intent in the search request. For example, slots include title, task, acceptance type, and others. The acceptance type represents how the user uses the search object; for example, acceptance types include listening to audiobooks, reading audiobooks, and reading comics. Others represent other preset types of slots.
[0036] A search request is the content entered into the search box. A search request can include a request to search for a novel, or a request to search for news articles, or a request to search for academic papers, etc.
[0037] Since slots represent identifiers of target content associated with the search intent within a search request, the search request can be divided into several content segments, known as slot fragments, based on the type of these identifiers. A slot fragment represents the content corresponding to each segment after the search request is divided into slots. For example, a search request like "aabbbee" can be divided into three slot fragments: aa, bbb, and ee. "aa" is labeled as a title, "bbb" as an author, and "ee" as a role.
[0038] For example, the search request is divided into slots and each slot segment is labeled with a type, including: dividing the search request into slots using a pre-trained slot model to obtain several slot segments, and labeling the type corresponding to each slot segment.
[0039] In this embodiment, a training sample set is constructed using the title, person, and acceptance type of the search object. Each training sample is labeled with a slot label corresponding to the title, person, and acceptance type of the search object, and a slot model is trained based on the training sample set. The person includes authors and roles, etc.
[0040] The search request is input into a pre-trained slot model, which performs character-level sequence labeling on the search request and outputs several slot segments with type labels, thus completing the division and type labeling of the search request.
[0041] S120. Perform intent recognition on the slot segment according to the type of each slot segment.
[0042] For example, for each slot segment, the target content of the slot segment is matched with a set cache according to the type of the slot segment to obtain a cache matching result.
[0043] If both the cache matching result and the type of the slot fragment meet the set conditions, the slot fragment is input into the pre-trained intent recognition model, and the intent recognition model is used to perform intent recognition on the slot fragment to obtain the model recognition result.
[0044] Based on the cache matching result and model recognition result corresponding to each slot segment, the intent recognition result corresponding to the slot segment is determined.
[0045] The defined cache includes the association between the attribute information of the search object and the genre of the search object. The attribute information of the search object may include at least one of the title, role, and author of the search object. In some embodiments, the defined cache may be constructed in key-value pairs, using at least one of the title, role, and author of the search object as the key and the genre of the search object as the value. For example, the defined cache may be constructed by associating at least one of the title, role, and author of each search object in all search objects entering the fine-sorting process with the genre of the search object.
[0046] In this embodiment of the disclosure, for each slot segment, the target content of the slot segment is matched with a set cache according to the type of the slot segment to obtain a cache matching result, including: for each slot segment, if the type of the slot segment is a title, the slot segment is matched with the titles of each search object in the set cache, and the cache matching result of the slot segment is determined according to the genre associated with the successfully matched title.
[0047] Alternatively, for each slot fragment, if the type of the slot fragment is a character, then the slot fragment is matched with the authors and / or characters of each search object in the set cache, and the cache matching result of the slot fragment is determined according to the genre associated with the successfully matched authors and / or characters.
[0048] By using a slot model to partition and label search requests by slot, different types of slot fragments can be cached separately, reducing the chances of complex search requests missing any cache.
[0049] In some embodiments, for each slot segment, if the number of cached matching results is less than a set threshold, and the type of the slot segment is non-person, then it is determined that both the cached matching results and the type of the slot segment meet the set conditions. The set threshold can be set according to actual needs, and it represents the case where the number of cached matching results is relatively small.
[0050] If both the cache matching result and the type of the slot fragment meet the set conditions, the slot fragment is input into the pre-trained intent recognition model, and the intent recognition model outputs the model recognition result corresponding to the slot fragment that meets the set conditions.
[0051] The intent recognition model is based on a training sample set consisting of the title, role, and author of historical search requests, and the training samples are labeled with the search intent of the historical search requests. The intent recognition model is trained through the training sample set.
[0052] If the type of the slot fragment is a person, then the slot fragment is not input into the intent recognition model to avoid the situation where the accuracy of intent recognition is affected by the person's name.
[0053] For search requests where the corresponding slot segment is not a person / entity, the intent recognition result for that slot segment is determined based on the cache matching result and the model recognition result. For search requests where the corresponding slot segment is a person / entity, the intent recognition result for that slot segment is determined based on the cache matching result.
[0054] S130. The search intent of the search request is obtained by fusing the intent recognition results corresponding to each slot segment.
[0055] In this embodiment of the disclosure, there may be duplicate intent recognition results between different slot segments. Based on the frequency of intent recognition results between different slot segments, the intent recognition results corresponding to all slot segments corresponding to the search request are merged to obtain the search intent of the search request.
[0056] For example, for each slot segment, the frequency of each intent recognition result corresponding to the slot segment is determined. Based on the frequency of each intent recognition result corresponding to each slot segment, the intent recognition results corresponding to each slot segment are fused to obtain the search intent of the search request.
[0057] For each slot segment, the frequency of the intent recognition result corresponding to the slot segment can be the number of times that intent recognition result appears repeatedly in the intent recognition results corresponding to other slots.
[0058] For each slot segment, the intent recognition result corresponding to each slot segment is compared with the intent recognition results corresponding to other slot segments corresponding to the search request, and the frequency of each intent recognition result corresponding to the slot segment is determined based on the comparison results.
[0059] If the frequency of cross-slot occurrence is used to represent the number of times each intent recognition result corresponding to the current slot segment appears repeatedly in the intent recognition results corresponding to other slots, then the frequency of cross-slot occurrence of each intent recognition result is counted, and only one result is retained for repeated intent recognition results, so as to realize the fusion of the intent recognition results corresponding to each slot segment and obtain the search intent corresponding to the search request.
[0060] The technical solution of this disclosure divides search requests into slots, labels each slot segment with a type, identifies the intent of each slot segment according to its type, and then fuses the intent identification results corresponding to each slot to obtain the search intent of the search request. This can increase the proportion of search requests that identify the search intent, and also improve the accuracy and recall of intent identification for complex search requests, thus solving the problem that the accuracy and recall of intent identification cannot meet search needs.
[0061] Figure 2 This is a flowchart illustrating another search intent recognition method provided by an embodiment of this disclosure. Based on the above embodiments, this disclosure further specifies the determination of the similarity between the search intent and the genre of the candidate search object. Figure 2 As shown, the method includes:
[0062] S210. The search request is divided into slots, and each slot segment is labeled with a type, wherein the slot represents the identifier of the target content in the search request that is associated with the search intent.
[0063] S220. Perform intent recognition on the slot segment according to the type of each slot segment.
[0064] S230. For each slot segment, determine the frequency of each intent recognition result corresponding to the slot segment.
[0065] S240. Based on the frequency of each intent recognition result corresponding to each slot segment, fuse the intent recognition results corresponding to each slot segment to obtain the search intent of the search request.
[0066] S250. Determine the weight of each search intent based on the frequency of each search intent corresponding to the search request.
[0067] For example, the highest frequency is determined based on the frequency of each search intent in the search request, and the weight of each search intent in the search request is determined based on the highest frequency and the frequency of each search intent.
[0068] In this embodiment of the disclosure, the frequency of each search intent in a search request can be normalized to obtain the weight of each search intent. For example, the most frequent search intent can be determined based on the frequency of each search intent in the search request, and the frequency corresponding to the most frequent search intent can be taken as the highest frequency. For each search intent corresponding to the search request, the weight of the search intent is determined based on the ratio of the frequency of the search intent to the highest frequency.
[0069] S260. Determine the intersection of the search intent corresponding to the search request and the genre of the candidate search object, and determine the similarity between the search intent and the genre of the candidate search object based on the weight corresponding to each search intent in the intersection.
[0070] The candidate search objects can be a set of search objects in the search request. For example, the candidate search objects could be the set of all search objects that have entered the fine-grained ranking. Optionally, the types of candidate search objects could include novels, news articles, and academic papers.
[0071] The intersection of the search intent corresponding to the search request and the genre of the candidate search objects can include one or more search intents in the search intent corresponding to the search request that have the same genre as the candidate search objects. For example, if the search intents include: aa, bb, cc, and the weight of search intent aa is 0.5, the weight of search intent bb is 0.2, and the weight of search intent cc is 0.3. For M candidate search objects (DOC1, DOC2, ..., DOC...) entering the fine sorting... M And the genre corresponding to each candidate search object can be represented as: DOC1: aa, ee; DOC2: aa, cc; ...; DOC M :bbb,ff. Therefore, the intersection of the search intent corresponding to the search request and the genre of DOC1 is aa; the intersection of the search intent corresponding to the search request and the genre of DOC2 is aa and cc; ...; the intersection of the search intent corresponding to the search request and the genre of DOC... M The intersection of the genres is bbb.
[0072] For each candidate search object DOC i 'i' represents the ID of the target search object, which is determined based on the search request and the DOC. i The numerator xi is determined by the weight of each search intent in the intersection of search intents. This is based on the search request and the DOC. i Given the maximum weight in the intersection of search intents and the length of the search intent corresponding to the search request, determine the denominator y. Then, the search intent and DOC... i The similarity of genres can be expressed as
[0073] For an example where the search intents include: aa, bb, cc, and the weight of search intent aa is 0.5, the weight of search intent bb is 0.2, and the weight of search intent cc is 0.3, the intersection of the search request and DOC1 with respect to the search intent is aa. Therefore, the numerator is the weight of aa, i.e., 0.5 as the numerator, and the denominator is... The similarity between search intent and the genre of DOC1 is represented as follows: The intersection of the search request and DOC2 with respect to the search intent is aa and cc. Therefore, the numerator is the sum of the weights of aa and cc, i.e., 0.8 as the numerator, and the denominator is... The similarity between search intent and the genre of DOC2 is represented as follows: The similarity between the search intent and the genre of the remaining search objects is calculated using a similarity-based approach.
[0074] Figure 3 This is a schematic diagram of a search intent recognition framework provided in an embodiment of this disclosure. Based on the above embodiments, this disclosure provides a specific process for recognizing the search intent corresponding to a search request. Figure 3 As shown, the framework includes: a slot partitioning module 310, a cache query module 320, a model recognition module 330, and an intent fusion module 340. The slot partitioning module 310 includes a slot model, which is used to partition and label search requests. When a search request is input into the slot partitioning module 310, the slot model performs character-level sequence labeling on the search request and outputs several slot fragments with type labels. The sequence labeling can be the process of adding labels to the character sequence corresponding to the search request. For example, if the search request is "aabbb", by performing sequence labeling on this search request, slot 1: title is obtained. <aa>Slot 2: Author <bbb>The slot fragments related to the search request are input into the cache query module 320. The cache query module 320 includes the association between the attribute information and genre of the candidate search objects, where the attribute information includes site title, role, external book, and author, etc. For each slot fragment, the cache query module 320 matches the target content of the slot fragment with the attribute information in the set cache according to the type of the slot fragment, and takes the genre associated with the successfully matched attribute information as the cache matching result. For the example of the search request "aabbb", the cache matching result may include the cache matching result of slot 1 and the cache matching result of slot 2. The cache matching result corresponding to each slot fragment is input into the model recognition module 330. The model recognition module 330 determines whether the number of non-duplicate results in all cache matching results corresponding to the search request is less than a set threshold. If so, the slot fragments that are not related to people in the slot fragments corresponding to the search request are input into the intent recognition model, and the intent recognition model outputs the model recognition result. For the example of the search request "aabbb", slot 1 is input into the intent recognition model, and the model recognition result of slot 1 is obtained from the intent recognition model output. Since slot 2's type tag is "author," slot 2 is not input into the intent recognition model. The cached matching results and model recognition results for non-personal slot segments corresponding to the search request are input into the intent fusion module 340, as are the cached matching results for person slot segments. The intent fusion module 340 merges the cached matching results and model recognition results for the same slot segment to obtain the intent recognition result for each non-personal slot segment. Optionally, during the merging process of the cached matching results and model recognition results for the same slot segment, only one result is retained for duplicate results, and frequency accumulation is not performed. The intent recognition results for each slot segment corresponding to the search request are merged, and the frequency of cross-slot occurrences for each intent recognition result is counted. This frequency is input into the weight model, which determines the weight of the intent recognition result based on this frequency. The intent fusion module 340 outputs several weighted search intents.
[0075] The technical solution of this disclosure uses a slot model to divide search requests into slots and label them by type. This allows different types of slot fragments to hit the cache separately, reducing the possibility of complex search requests failing to hit any cache. Furthermore, based on the weight of each search intent in the intersection of the search request and the candidate search object with respect to the search intent, the similarity between the search intent and the candidate search object's genre is determined. The search object to be recalled is then determined based on the similarity, significantly improving the recall effect, enhancing user satisfaction with search intent, and improving the user search experience.
[0076] Figure 4 This is a schematic diagram of a search intent recognition device provided in an embodiment of this disclosure. This device can execute any of the search intent recognition methods provided in this disclosure, and can be implemented in software and / or hardware. Optionally, it can be implemented using an electronic device, such as a mobile terminal, a PC, or a server. Figure 4 As shown, the device includes: a slot division module 410, an intent recognition model 420, and a result fusion module 430.
[0077] The slot partitioning module 410 is used to partition the search request into slots and label each slot segment with a type, wherein the slot represents the identifier of the target content in the search request that is associated with the search intent.
[0078] Intent recognition model 420 is used to perform intent recognition on the slot segment according to the type of each slot segment;
[0079] The result fusion module 430 is used to fuse the intent recognition results corresponding to each slot segment to obtain the search intent of the search request.
[0080] The technical solution provided in this disclosure divides search requests into slots, labels each slot segment with a type, identifies the intent of each slot segment according to its type, and then fuses the intent identification results corresponding to each slot to obtain the search intent of the search request. This can increase the proportion of search requests that identify the search intent, and also improve the accuracy and recall of intent identification for complex search requests, thus solving the problem that the accuracy and recall of intent identification cannot meet search needs.
[0081] Optionally, the slot partitioning module 410 is specifically used for:
[0082] The search request is divided into slots using a pre-trained slot model, resulting in several slot segments, and the type corresponding to each slot segment is labeled.
[0083] Optionally, the intent recognition model 420 is specifically used for:
[0084] For each slot segment, the target content of the slot segment is matched with the set cache according to the type of the slot segment to obtain the cache matching result. The set cache includes the attribute information of the search object and the association relationship between the genre of the search object.
[0085] If both the cache matching result and the type of the slot fragment meet the set conditions, the slot fragment is input into the pre-trained intent recognition model, and the intent recognition model is used to perform intent recognition on the slot fragment to obtain the model recognition result.
[0086] Based on the cache matching result and model recognition result corresponding to each slot segment, the intent recognition result corresponding to the slot segment is determined.
[0087] Optionally, the result fusion module 430 includes:
[0088] The frequency determination unit is used to determine the frequency of each intent recognition result corresponding to each slot segment for each slot segment;
[0089] The search intent determination unit is used to fuse the intent recognition results corresponding to each slot segment based on the frequency of each intent recognition result corresponding to each slot segment to obtain the search intent of the search request.
[0090] Furthermore, the frequency determination unit is specifically used for:
[0091] For each slot segment, the intent recognition result corresponding to each slot segment is compared with the intent recognition results corresponding to other slot segments corresponding to the search request, and the frequency of each intent recognition result corresponding to the slot segment is determined based on the comparison results.
[0092] Optionally, the device further includes:
[0093] The weight determination module is used to determine the weight of each search intent based on the frequency of each intent recognition result corresponding to each slot segment after fusing the intent recognition results corresponding to each slot segment to obtain the search intent of the search request.
[0094] The similarity determination module is used to determine the intersection of the search intent corresponding to the search request and the genre of the candidate search object, and to determine the similarity between the search intent and the genre of the candidate search object based on the weight corresponding to each search intent in the intersection.
[0095] Optionally, the weight determination module is specifically used for:
[0096] The highest frequency is determined based on the frequency of each search intent in the search request, and the weight of each search intent in the search request is determined based on the highest frequency and the frequency of each search intent.
[0097] The search intent recognition device provided in this disclosure can execute the search intent recognition method provided in any embodiment of this disclosure, and has the corresponding functional modules and beneficial effects of executing the method.
[0098] It is worth noting that the various units and modules included in the above-mentioned device are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the protection scope of the embodiments of this disclosure.
[0099] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. Reference is made below. Figure 5 It illustrates an electronic device suitable for implementing embodiments of the present disclosure (e.g., Figure 5 The diagram below shows the structure of the terminal device or server 500. The terminal device in this embodiment may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), and vehicle terminals (e.g., vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 5 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.
[0100] like Figure 5 As shown, electronic device 500 may include a processing unit (e.g., central processing unit, graphics processor, etc.) 501, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 502 or a program loaded from storage device 508 into random access memory (RAM) 503. The RAM 503 also stores various programs and data required for the operation of electronic device 500. The processing unit 501, ROM 502, and RAM 503 are interconnected via bus 504. An edit / output (I / O) interface 505 is also connected to bus 504.
[0101] Typically, the following devices can be connected to I / O interface 505: input devices 506 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 507 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 508 including, for example, magnetic tapes, hard disks, etc.; and communication devices 509. Communication device 509 allows electronic device 500 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 5 An electronic device 500 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively.
[0102] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this 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 methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 509, or installed from a storage device 508, or installed from a ROM 502. When the computer program is executed by the processing device 501, it performs the functions defined in the methods of embodiments of this disclosure.
[0103] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.
[0104] The electronic device provided in this embodiment and the search intent recognition method provided in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.
[0105] This disclosure provides a computer storage medium storing a computer program that, when executed by a processor, implements the search intent recognition method provided in the above embodiments.
[0106] It should be noted that the computer-readable medium described in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.
[0107] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol, such as HTTP (Hypertext Transfer Protocol), and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks ("LANs"), wide area networks ("WANs"), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.
[0108] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.
[0109] The aforementioned computer-readable medium carries one or more programs that, when executed by the electronic device, cause the electronic device to:
[0110] The search request is divided into slots, and each slot segment is labeled with a type, wherein the slot represents the identifier of the target content in the search request that is associated with the search intent;
[0111] Intent recognition is performed on the slot segments based on the type of each slot segment;
[0112] The search intent of the search request is obtained by fusing the intent recognition results corresponding to each slot segment.
[0113] Computer program code for performing the operations of this disclosure can be written in one or more programming languages or a combination thereof, including but not limited to object-oriented programming languages such as Java, Smalltalk, and C++, as well as conventional procedural programming languages such as "C" or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).
[0114] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.
[0115] The units described in the embodiments of this disclosure can be implemented in software or hardware. The names of the units are not, in some cases, intended to limit the specific unit.
[0116] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: Field Programmable Gate Arrays (FPGAs), Application-Specific Integrated Circuits (ASICs), Application Standard Products (ASSPs), System-on-Chip (SoCs), Complex Programmable Logic Devices (CPLDs), and so on.
[0117] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.
[0118] According to one or more embodiments of this disclosure, Example 1 provides a search intent recognition method, including:
[0119] The search request is divided into slots, and each slot segment is labeled with a type, wherein the slot represents the identifier of the target content in the search request that is associated with the search intent;
[0120] Intent recognition is performed on the slot segments based on the type of each slot segment;
[0121] The search intent of the search request is obtained by fusing the intent recognition results corresponding to each slot segment.
[0122] According to one or more embodiments of this disclosure, Example 2, based on the method described in Example 1, includes the following: Slotting the search request and labeling each slot segment with a type.
[0123] The search request is divided into slots using a pre-trained slot model, resulting in several slot segments, and the type corresponding to each slot segment is labeled.
[0124] According to one or more embodiments of this disclosure, Example 3, based on the method of Example 1, includes performing intent recognition on the slot segment according to the type of each slot segment, comprising:
[0125] For each slot segment, the target content of the slot segment is matched with the set cache according to the type of the slot segment to obtain the cache matching result. The set cache includes the attribute information of the search object and the association relationship between the genre of the search object.
[0126] If both the cache matching result and the type of the slot fragment meet the set conditions, the slot fragment is input into the pre-trained intent recognition model, and the intent recognition model is used to perform intent recognition on the slot fragment to obtain the model recognition result.
[0127] Based on the cache matching result and model recognition result corresponding to each slot segment, the intent recognition result corresponding to the slot segment is determined.
[0128] According to one or more embodiments of this disclosure, Example 4, based on the method described in Example 1, involves fusing the intent recognition results corresponding to each slot segment to obtain the search intent of the search request, including:
[0129] For each slot segment, determine the frequency of each intent recognition result corresponding to the slot segment;
[0130] Based on the frequency of each intent recognition result corresponding to each slot segment, the intent recognition results corresponding to each slot segment are fused to obtain the search intent of the search request.
[0131] According to one or more embodiments of this disclosure, Example 5 describes the method described in Example 4, wherein determining the frequency of each intent recognition result corresponding to each slot segment includes:
[0132] For each slot segment, the intent recognition result corresponding to each slot segment is compared with the intent recognition results corresponding to other slot segments corresponding to the search request, and the frequency of each intent recognition result corresponding to the slot segment is determined based on the comparison results.
[0133] According to one or more embodiments of this disclosure, Example 6, based on the method described in Example 4, after fusing the intent recognition results corresponding to each slot segment according to the frequency of each intent recognition result corresponding to each slot segment to obtain the search intent of the search request, further includes:
[0134] The weight of each search intent is determined based on the frequency of each search intent corresponding to the search request;
[0135] Determine the intersection of the search intent corresponding to the search request and the genre of the candidate search object, and determine the similarity between the search intent and the genre of the candidate search object based on the weight corresponding to each search intent in the intersection.
[0136] According to one or more embodiments of this disclosure, Example 7 describes the method described in Example 6, wherein determining the weight of each search intent based on the frequency of each search intent corresponding to the search request includes:
[0137] The highest frequency is determined based on the frequency of each search intent in the search request, and the weight of each search intent in the search request is determined based on the highest frequency and the frequency of each search intent.
[0138] According to one or more embodiments of this disclosure, Example 8 provides a search intent recognition device, including:
[0139] The slot partitioning module is used to partition the search request into slots and label each slot segment with a type. The slot represents the identifier of the target content in the search request that is associated with the search intent.
[0140] An intent recognition model is used to recognize the intent of each slot segment based on its type.
[0141] The result fusion module is used to fuse the intent recognition results corresponding to each slot segment to obtain the search intent of the search request.
[0142] According to one or more embodiments of this disclosure, Example 9 provides an electronic device, the electronic device comprising:
[0143] One or more processors;
[0144] Storage device for storing one or more programs.
[0145] When the one or more programs are executed by the one or more processors, the one or more processors implement the search intent recognition method as described in any of Examples 1-7.
[0146] According to one or more embodiments of the present disclosure, Example 10 provides a storage medium containing computer-executable instructions that, when executed by a computer processor, are used to perform a search intent recognition method as described in any of Examples 1-7.
[0147] The above description is merely a preferred embodiment of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features disclosed in this disclosure that have similar functions.
[0148] Furthermore, while the operations are described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or in a sequential order. In certain environments, multitasking and parallel processing may be advantageous. Similarly, while several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of this disclosure. Certain features described in the context of individual embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented individually or in any suitable sub-combination in multiple embodiments.
[0149] Although the subject matter has been described using language specific to structural features and / or methodological logic, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are merely illustrative examples of implementing the claims.< / bbb> < / aa>
Claims
1. A search intent recognition method characterized by, include: The search request is divided into slots, and each slot segment is labeled with a type, wherein the slot represents the identifier of the target content in the search request that is associated with the search intent; For each slot segment, the target content of the slot segment is matched with the set cache according to the type of the slot segment to obtain the cache matching result. The set cache includes the attribute information of the search object and the association relationship between the genre of the search object. If both the cache matching result and the type of the slot fragment meet the set conditions, then the slot fragment is input into a pre-trained intent recognition model, and the intent recognition model is used to perform intent recognition on the slot fragment to obtain the model recognition result. The condition that both the cache matching result and the type of the slot fragment meet the set conditions includes: if the number of cache matching results is less than a set threshold, and the type of the slot fragment is non-human, then it is determined that both the cache matching result and the type of the slot fragment meet the set conditions. Based on the cache matching result and model recognition result corresponding to each slot segment, the intent recognition result corresponding to the slot segment is determined; The search intent of the search request is obtained by fusing the intent recognition results corresponding to each slot segment.
2. The method of claim 1, wherein, The process of dividing search requests into slots and labeling each slot segment with its type includes: The search request is divided into slots using a pre-trained slot model, resulting in several slot segments, and the type corresponding to each slot segment is labeled.
3. The method according to claim 1, characterized in that, The search intent of the search request is obtained by fusing the intent recognition results corresponding to each slot segment, including: For each slot segment, determine the frequency of each intent recognition result corresponding to the slot segment; Based on the frequency of each intent recognition result corresponding to each slot segment, the intent recognition results corresponding to each slot segment are fused to obtain the search intent of the search request.
4. The method according to claim 3, characterized in that, For each slot segment, determining the frequency of each intent recognition result corresponding to the slot segment includes: For each slot segment, the intent recognition result corresponding to each slot segment is compared with the intent recognition results corresponding to other slot segments corresponding to the search request, and the frequency of each intent recognition result corresponding to the slot segment is determined based on the comparison results.
5. The method according to claim 3, characterized in that, After fusing the intent recognition results corresponding to each slot segment based on the frequency of each intent recognition result corresponding to each slot segment to obtain the search intent of the search request, the method further includes: The weight of each search intent is determined based on the frequency of each search intent corresponding to the search request; Determine the intersection of the search intent corresponding to the search request and the genre of the candidate search object, and determine the similarity between the search intent and the genre of the candidate search object based on the weight corresponding to each search intent in the intersection.
6. The method according to claim 5, characterized in that, The step of determining the weight of each search intent based on the frequency of each search intent corresponding to the search request includes: The highest frequency is determined based on the frequency of each search intent in the search request, and the weight of each search intent in the search request is determined based on the highest frequency and the frequency of each search intent.
7. A search intent recognition device, characterized in that, include: The slot partitioning module is used to partition the search request into slots and label each slot segment with a type. The slot represents the identifier of the target content in the search request that is associated with the search intent. An intent recognition model is used to recognize the intent of each slot segment based on its type. The result fusion module is used to fuse the intent recognition results corresponding to each slot segment to obtain the search intent of the search request; The intent recognition model is specifically used for: For each slot segment, the target content of the slot segment is matched with the set cache according to the type of the slot segment to obtain the cache matching result. The set cache includes the attribute information of the search object and the association relationship between the genre of the search object. If both the cache matching result and the type of the slot fragment meet the set conditions, then the slot fragment is input into a pre-trained intent recognition model, and the intent recognition model is used to perform intent recognition on the slot fragment to obtain the model recognition result. The condition that both the cache matching result and the type of the slot fragment meet the set conditions includes: if the number of cache matching results is less than a set threshold, and the type of the slot fragment is non-human, then it is determined that both the cache matching result and the type of the slot fragment meet the set conditions. Based on the cache matching result and model recognition result corresponding to each slot segment, the intent recognition result corresponding to the slot segment is determined.
8. An electronic device, characterized in that, The electronic device includes: One or more processors; Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the search intent recognition method as described in any one of claims 1-6.
9. A storage medium containing computer-executable instructions, characterized in that, The computer-executable instructions, when executed by a computer processor, are used to perform the search intent recognition method as described in any one of claims 1-6.