Search term identification method and apparatus, electronic device, and storage medium

Abstract

The present disclosure provides a search term recognition method and device, electronic equipment and storage medium, relates to the technical field of data processing, in particular to the technical field of cloud computing and intelligent search, and includes: obtaining a search term input by a user; identifying the search term by using a first search term recognition model to obtain a search term recognition result and a search term identifier; determining whether the search term is of a first specific type according to the search term recognition result; in response to the search term being of the first specific type, matching the search term identifier in a first preset database; and in response to a successful match, obtaining a matching result corresponding to the search term and sending the matching result to the user.

Description

Technical Field

[0001] This disclosure relates to the field of data processing technology, and in particular to the fields of cloud computing and intelligent search technology. Background Technology

[0002] Currently, there are two methods for identifying search terms: one is manual identification, which requires a lot of manpower, and the other is to use machine algorithms to identify search terms through statistical features or based on language models. However, the accuracy of this method in identifying search terms is not high. Summary of the Invention

[0003] This disclosure provides a method, apparatus, electronic device, and storage medium for identifying search terms.

[0004] According to one aspect of this disclosure, a method for identifying search terms is provided, comprising:

[0005] Get the search terms entered by the user;

[0006] The search terms are identified using the first search term recognition model to obtain search term recognition results and search term identifiers;

[0007] Based on the search term identification results, determine whether the search term belongs to the first specific type;

[0008] In response to the search term being of a first specific type, a match is made from a first preset database based on the search term identifier;

[0009] In response to a successful match, the matching result corresponding to the search term is obtained and sent to the user.

[0010] According to another aspect of this disclosure, a search term recognition device is provided, comprising:

[0011] The data acquisition module is used to obtain the search terms input by the user.

[0012] The recognition module is used to recognize the search terms using the first search term recognition model to obtain the search term recognition result and the search term identifier;

[0013] The processing module is used to determine whether the search term belongs to a first specific type based on the search term recognition result;

[0014] The matching module is configured to, in response to the search term being of a first specific type, match from a first preset database based on the search term identifier;

[0015] The communication module is used to obtain the matching result corresponding to the search term and send the matching result to the user in response to a successful match.

[0016] According to another aspect of this disclosure, an electronic device is provided, comprising:

[0017] At least one processor; and

[0018] A memory communicatively connected to the at least one processor; wherein,

[0019] The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform any of the methods described above.

[0020] According to another aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions, wherein the computer instructions are configured to cause the computer to perform the method described in any of the preceding claims.

[0021] According to another aspect of this disclosure, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the method described in any of the preceding claims.

[0022] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0023] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein:

[0024] Figure 1 This is a flowchart illustrating the search term recognition method provided according to embodiments of this disclosure;

[0025] Figure 2 This is a schematic diagram of a first specific type of search term provided according to an embodiment of this disclosure;

[0026] Figure 3 This is a schematic diagram of the structure of the second and third search term recognition models provided according to embodiments of this disclosure;

[0027] Figure 4 This is a schematic diagram of a second specific type of search term provided according to an embodiment of this disclosure;

[0028] Figure 5 This is a schematic diagram of the structure of the search term recognition device provided according to an embodiment of the present disclosure;

[0029] Figure 6 This is a block diagram of an electronic device used to implement the search term recognition method of the embodiments of this disclosure. Detailed Implementation

[0030] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0031] To improve the recognition of search terms and reduce labor costs, such as Figure 1 As shown, one embodiment of this disclosure provides a method for identifying search terms, the method comprising:

[0032] Step 101: Obtain the search terms entered by the user.

[0033] Get the search terms that users enter in a search engine or any search input box.

[0034] Step 102: Use the first search term recognition model to recognize the search term and obtain the search term recognition result and search term identifier.

[0035] The search term is identified using a pre-trained first search term recognition model to obtain the recognition result and search term identifier.

[0036] Search term identifiers can be keywords, feature data, or any identifier that can represent the characteristics of the search term.

[0037] Step 103: Determine whether the search term is a first specific type based on the search term recognition result.

[0038] Based on the recognition results obtained after the first search term recognition model identifies the search term, it is determined whether the search term belongs to the first specific type.

[0039] like Figure 2 As shown, the first specific type refers to whether the search term is a type of internet slang, dialect, professional term, or ordinary entity. Internet slang refers to words, phrases, sentences, etc., that are entirely generated online. Dialect refers to search terms that are common expressions in non-Mandarin, generally belonging only to a certain region or having a unique meaning. Professional term refers to search terms that are professional content in a certain field, belonging to the derivative content of entity words. Ordinary entity refers to the meaning and definition of specific entities, which are mostly objectively existing and distinguishable things.

[0040] Step 104: In response to the search term being a first specific type, a match is performed from a first preset database based on the search term identifier.

[0041] When a search term is determined to be a first specific type, it is matched from a first preset database based on the search term identifier;

[0042] The first preset database contains multiple terms and corresponding search term identifiers for each term. Each term has a preset search term definition. For a first specific type of search term, the matching results obtained after a successful match contain at least one term. The content of the term can better meet the user's search needs and improve the user experience.

[0043] Step 105: In response to a successful match, obtain the matching result corresponding to the search term and send the matching result to the user.

[0044] If a match is successful, the matching result obtained from the search term will be sent to the user.

[0045] By using a pre-trained first search term recognition model to identify search terms, search term recognition results and search term identifiers are obtained. Then, based on the search term recognition results, it is determined whether the search term belongs to a first specific type. When the search term is determined to be the first specific type, it is matched from a first preset database based on the search term identifier. Finally, the matching result is sent to the user. After determining that the search term is the first specific type, the matching result is matched from the first preset database based on the search term identifier. This method can effectively improve the accuracy of search term recognition, and the recognition process does not require manual intervention, thus reducing a lot of labor costs.

[0046] In step 101, before obtaining the search terms input by the user, in one possible implementation, the method further includes:

[0047] Step 201: Collect the first training dataset.

[0048] Collect the first training dataset, which contains multiple training data sets.

[0049] Step 202: Use the second search term recognition model to identify multiple training data in the first training dataset to obtain the labels corresponding to each training data.

[0050] The pre-trained second search term recognition model is used to identify multiple training data in the first training dataset to obtain the labels corresponding to each training data.

[0051] Step 203: Perform model distillation on the second search term recognition model to obtain the third search term recognition model.

[0052] The second search term recognition model is distilled to obtain the third search term recognition model.

[0053] The number of Transformer modules in the second search term recognition model is reduced to obtain the third search term recognition model. The input of the second search term recognition model is character-level data, while the input of the third search term recognition model is word-level data.

[0054] like Figure 3 As shown, Figure 3 The left side shows the structure of the second search term recognition model, and the right side shows the structure of the third search term recognition model. The Transformer module in the second search term recognition model has 12 layers, while the simplified third search term recognition model has only 3 layers in its Transformer module, which greatly reduces the time required for data processing and improves the efficiency of search term recognition. Figure 3 The data input to the second search term recognition model is {9, 9, 9, 7, 7, is, what, what, meme}, while the data input to the third search term recognition model is {99977, is, what, meme}.

[0055] Step 204: Train the third search term recognition model using the labeled first training dataset to obtain the first search term recognition model.

[0056] The third search term recognition model is trained using the labeled first training dataset to obtain the first search term recognition model;

[0057] The labels of each training data in the first training dataset are obtained by the second search term recognition model. Therefore, the first search term recognition model obtained by training the third search term recognition model using the labeled first training dataset can obtain the search term recognition ability of the second search term recognition model, which greatly improves the search term recognition ability of the first search term recognition model.

[0058] In step 201, before collecting the first training dataset, in one possible implementation, the method further includes:

[0059] Step 301: Collect the second training dataset, where each training data point has a corresponding label.

[0060] Collect a second training dataset, in which each training data point has its own label.

[0061] Step 302: Train the language model using the second training dataset to obtain the second retrieval word recognition model.

[0062] The language model was trained using the second training dataset to obtain the second retrieval word recognition model;

[0063] The language model is trained using a second training dataset with built-in labels, so that the resulting second search term recognition model has the ability to accurately recognize search terms;

[0064] In this embodiment, the language model can be a pre-trained language model based on the text center.

[0065] In step 103, after determining whether the search term belongs to the first specific type based on the search term identification result, in one possible implementation, the method further includes:

[0066] Step 401: In response to the search term being a second specific type, search results are matched from a preset database based on the search term.

[0067] When a search term is determined to be a second specific type, search results are matched from a pre-defined database based on the search term.

[0068] The pre-defined database contains multiple search results and the corresponding search terms for each search result;

[0069] like Figure 4 As shown, the second specific type refers to whether the search term is a multi-entity category, an ambiguous category, a language explanation category, a translation category, a method category, a pornographic / anti-pornographic category, a medical category, a music category, a personal name category, or a food category. A multi-entity category means the search term contains multiple entities; an ambiguous category means the entity meaning in the search term is unclear; a language explanation category means the search term is a word, idiom, classical saying, proverb, allegorical saying, or a sentence from a literary work; a translation category means the search term is an explanation of a foreign language word; a method category means the search term is an experiential need, containing words like "how to do it" or "why"; a pornographic / anti-pornographic category means the search term contains negative information; a medical category means the search term is a description of the efficacy and effects of medicines; a music category means the search term contains music-related content; a personal name category means the search term is a person's real name, stage name, or nickname; and a food category means the search term contains food-related content.

[0070] Step 402: In response to a successful match, obtain the matching result corresponding to the search term and send the matching result to the user.

[0071] When a search term is identified as a second specific type, it means that there is no corresponding entry for that search term. In this case, a match is performed from the second preset database and the matching result is sent to the user, which can improve the stability and usability of the identification.

[0072] In step 204, after obtaining the first search term recognition model, in one possible implementation, the method further includes:

[0073] Step 501: Collect the third training dataset.

[0074] The third training dataset contains multiple training data sets and corresponding labels for each training data set.

[0075] Step 502: Divide the third training dataset into multiple fourth training datasets.

[0076] The third training dataset is divided into multiple fourth training datasets;

[0077] For example, the third training dataset is randomly divided into five fourth training datasets.

[0078] Step 503: Train multiple first search term recognition models using the multiple fourth training datasets to obtain multiple trained first search term recognition models.

[0079] Multiple first-search-term recognition models were trained using multiple fourth-level training datasets to obtain multiple trained first-search-term recognition models.

[0080] For example, five first-word recognition models were trained using five fourth training datasets, resulting in first-word recognition model A, first-word recognition model B, first-word recognition model C, first-word recognition model D, and first-word recognition model E. The five first-word recognition models have different abilities to recognize search terms, so they need to be tested.

[0081] Step 504: Use the multiple trained first search term recognition models to identify the third training dataset and obtain the recognition result.

[0082] The third training dataset was identified using multiple trained first-word recognition models to obtain the recognition results;

[0083] For example, using five trained first search term recognition models A, B, C, D, and E to identify the third training dataset, the recognition results are A, B, C, D, and E, respectively.

[0084] Step 505: Determine the first search term recognition model with the best recognition effect based on the recognition results.

[0085] Based on the recognition results, determine the first search term recognition model that has the best recognition performance.

[0086] By training the first search term recognition model on multiple different fourth training datasets, and then validating the trained first search term recognition models on the third training dataset, the model with the highest retrieval accuracy and best performance can be determined, significantly improving the ability to recognize search terms.

[0087] In step 104, after matching from the first preset database based on the search term identifier, in one possible implementation, the method further includes:

[0088] When a search term fails to match data in the first preset database, it means that there is no corresponding entry for that search term. In this case, the search term is sent to the human department, which will create the corresponding entry so that subsequent users can find the entry for that search term and improve the user experience.

[0089] One embodiment of this disclosure provides a search term recognition device, such as... Figure 5 As shown, the device includes:

[0090] The acquisition module 10 is used to acquire the search terms input by the user;

[0091] The identification module 20 is used to identify the search terms using the first search term identification model to obtain the search term identification result and the search term identifier;

[0092] Processing module 30 is used to determine whether the search term is a first specific type based on the search term recognition result;

[0093] Matching module 40 is configured to, in response to the search term being of a first specific type, perform matching from a first preset database based on the search term identifier;

[0094] The communication module 50 is used to obtain the matching result corresponding to the search term and send the matching result to the user in response to a successful match.

[0095] The acquisition module 10 is also used to acquire the first training dataset;

[0096] The recognition module 20 is further configured to use the second search term recognition model to recognize multiple training data in the first training dataset and obtain labels corresponding to each training data.

[0097] The processing module 30 is further configured to perform model distillation on the second search term recognition model to obtain a third search term recognition model;

[0098] The processing module 30 is further configured to train the third search term recognition model using a labeled first training dataset to obtain a first search term recognition model.

[0099] The acquisition module 10 is also used to acquire a second training dataset, wherein each training data in the second training dataset has a corresponding label.

[0100] The processing module 30 is further configured to train the language model using the second training dataset to obtain a second search term recognition model.

[0101] The matching module 40 is further configured to, in response to the search term being a second specific type, perform matching from a second preset database based on the search term;

[0102] The communication module 50 is also used to obtain the matching result corresponding to the search term and send the matching result to the user in response to a successful match.

[0103] The acquisition module 10 is also used to acquire a third training dataset;

[0104] The processing module 30 is further configured to divide the third training dataset into multiple fourth training datasets;

[0105] The processing module 30 is further configured to train multiple first search term recognition models using the multiple fourth training datasets to obtain multiple trained first search term recognition models.

[0106] The recognition module 20 is also used to recognize the third training dataset using the multiple trained first search term recognition models to obtain recognition results;

[0107] The processing module 30 is further configured to determine the first search term recognition model with the best recognition effect based on the recognition result.

[0108] The acquisition, storage, and application of user personal information involved in the technical solution disclosed herein comply with relevant laws and regulations and do not violate public order and good morals.

[0109] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0110] Figure 6A schematic block diagram of an example electronic device 600 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0111] like Figure 6 As shown, device 600 includes a computing unit 601, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 602 or a computer program loaded from storage unit 608 into random access memory (RAM) 603. RAM 603 may also store various programs and data required for the operation of device 600. The computing unit 601, ROM 602, and RAM 603 are interconnected via bus 604. Input / output (I / O) interface 605 is also connected to bus 604.

[0112] Multiple components in device 600 are connected to I / O interface 605, including: input unit 606, such as keyboard, mouse, etc.; output unit 607, such as various types of monitors, speakers, etc.; storage unit 608, such as disk, optical disk, etc.; and communication unit 609, such as network card, modem, wireless transceiver, etc. Communication unit 609 allows device 600 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0113] The computing unit 601 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 601 performs the various methods and processes described above, such as the search term recognition method. For example, in some embodiments, the search term recognition method may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 608. In some embodiments, part or all of the computer program may be loaded and / or installed on device 600 via ROM 602 and / or communication unit 609. When the computer program is loaded into RAM 603 and executed by the computing unit 601, one or more steps of the search term recognition method described above may be performed. Alternatively, in other embodiments, the computing unit 601 may be configured to perform the search term recognition method by any other suitable means (e.g., by means of firmware).

[0114] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0115] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0116] 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.

[0117] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0118] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0119] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0120] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0121] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method for identifying search terms, comprising: Get the search terms entered by the user; The search terms are identified using the first search term recognition model to obtain search term recognition results and search term identifiers; Based on the search term identification results, determine whether the search term belongs to the first specific type; In response to the search term being of a first specific type, a match is performed from a first preset database based on the search term identifier; in response to a successful match, the matching result corresponding to the search term is obtained and the matching result is sent to the user; In response to the search term being a second specific type, a match is performed from a second preset database based on the search term; in response to a successful match, the matching result corresponding to the search term is obtained and the matching result is sent to the user; Before obtaining the user-input search terms, the process also includes: Collect the first training dataset; The second search term recognition model is used to identify multiple training data in the first training dataset to obtain the labels corresponding to each training data. The second search term recognition model is subjected to model distillation, and the Transformer module in the second search term recognition model is simplified from 12 layers to 3 layers to obtain the third search term recognition model. The input of the third search term recognition model is word-level data. The third search term recognition model is trained using the labeled first training dataset to obtain the first search term recognition model.

2. The method according to claim 1, further comprising, before collecting the first training dataset: Collect a second training dataset, where each training data point has a corresponding label; The language model is trained using the second training dataset to obtain the second search term recognition model.

3. The method according to claim 1, further comprising, after obtaining the first search term recognition model: Collect a third training dataset; The third training dataset is divided into multiple fourth training datasets; Multiple first search term recognition models are trained using the multiple fourth training datasets to obtain multiple trained first search term recognition models; The third training dataset is identified using the multiple trained first search term recognition models to obtain the recognition results; Based on the recognition results, determine the first search term recognition model with the best recognition performance.

4. A search term recognition device, comprising: The data acquisition module is used to obtain the search terms input by the user. The recognition module is used to recognize the search terms using the first search term recognition model to obtain the search term recognition result and the search term identifier; The processing module is used to determine whether the search term belongs to a first specific type based on the search term recognition result; The matching module is configured to, in response to the search term being of a first specific type, match from a first preset database based on the search term identifier; The communication module is used to obtain the matching result corresponding to the search term and send the matching result to the user in response to a successful match. The matching module is further configured to, in response to the search term being a second specific type, perform matching from a second preset database based on the search term; The communication module is also used to obtain the matching result corresponding to the search term and send the matching result to the user in response to a successful match; The acquisition module is also used to acquire the first training dataset; The recognition module is also used to recognize multiple training data in the first training dataset using the second search term recognition model to obtain labels corresponding to each training data. The processing module is also used to perform model distillation on the second search term recognition model, reducing the Transformer module in the second search term recognition model from 12 layers to 3 layers to obtain a third search term recognition model. The input of the third search term recognition model is word-level data. The processing module is further configured to train the third search term recognition model using a labeled first training dataset to obtain a first search term recognition model.

5. The apparatus according to claim 4, comprising: The acquisition module is also used to acquire a second training dataset, wherein each training data in the second training dataset has a corresponding label; The processing module is further configured to train the language model using the second training dataset to obtain a second search term recognition model.

6. The apparatus according to claim 4, comprising: The acquisition module is also used to acquire a third training dataset; The processing module is further configured to divide the third training dataset into multiple fourth training datasets; The processing module is also used to train multiple first search term recognition models using the multiple fourth training datasets to obtain multiple trained first search term recognition models. The recognition module is also used to recognize the third training dataset using the multiple trained first search term recognition models to obtain recognition results; The processing module is also used to determine the first search term recognition model with the best recognition effect based on the recognition result.

7. An electronic device, comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-3.

8. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-3.

9. A computer program product comprising a computer program that, when executed by a processor, implements the method according to any one of claims 1-3.

Images