CN117349515A - Search processing methods, electronic devices and storage media - Google Patents

Abstract

This application discloses a search processing method, electronic equipment and storage medium, and relates to the fields of large model technology and computer technology. The method includes: obtaining a search request; performing attribute feature analysis on the search request, and generating auxiliary information associated with the search request, where the auxiliary information is at least one target attribute feature of preset industry knowledge corresponding to the search request; based on the search request and Auxiliary information to obtain the search results corresponding to the search request. This application solves the technical problem in related technologies that it is difficult to inject industry knowledge into the search processing process, resulting in low search accuracy.

Description

Search processing methods, electronic devices and storage media

Technical field

This application relates to the fields of large model technology and computer technology, specifically, to a search processing method, electronic equipment and storage media.

Background technique

In recent years, search engine technology based on artificial intelligence models has been increasingly applied to all walks of life. However, traditional search engines are usually implemented based on small models and simple statistical algorithms, which are difficult to apply to complex scenarios. Although there are currently some search engines based on large models (such as Differential Search Index, DSI), existing search engines based on large models often rely entirely on the performance of large models, and it is difficult to integrate large models with traditional search engines. The combination of industry customization and manual intervention characteristics limits the search accuracy of existing large model-based search engines.

In response to the above problems, no effective solution has yet been proposed.

Contents of the invention

Embodiments of the present application provide a search processing method, electronic device, and storage medium to at least solve the technical problem in the related art that it is difficult to inject industry knowledge into the search processing process, resulting in low search accuracy.

According to one aspect of the embodiment of the present application, a search processing method is provided, including: obtaining a search request; performing attribute feature analysis on the search request, and generating auxiliary information associated with the search request, where the auxiliary information is a preset corresponding to the search request At least one target attribute feature of industry knowledge; based on the search request and auxiliary information, obtain the search results corresponding to the search request.

According to another aspect of the embodiment of the present application, another search processing method is also provided, including: obtaining a search request; using a differential search index model to perform attribute feature analysis on the search request to generate auxiliary information associated with the search request; and Request industry knowledge reasoning with the auxiliary information to output a target document identifier, where the auxiliary information is at least one target attribute feature of the preset industry knowledge corresponding to the search request; and obtain the search results corresponding to the search request based on the target document identifier.

According to another aspect of the embodiment of the present application, another search processing method is also provided, including: obtaining an e-commerce service search request; performing e-commerce attribute feature analysis on the e-commerce service search request, and generating an e-commerce service search request associated with the e-commerce service search request. E-commerce service auxiliary information, where the e-commerce service auxiliary information is the target e-commerce attribute feature combination of the e-commerce service industry knowledge corresponding to the e-commerce service search request; based on the e-commerce service search request and the e-commerce service auxiliary information, the e-commerce service auxiliary information is obtained E-commerce service search results corresponding to the service search request.

According to another aspect of the embodiment of the present application, an electronic device is also provided, including: a memory storing an executable program; a processor configured to run the program, wherein any one of the above search processing methods is executed when the program is running. .

According to another aspect of the embodiment of the present application, a computer-readable storage medium is also provided. The computer-readable storage medium includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute any arbitrary One of the above search processing methods.

In the embodiment of the present application, the search request is first obtained, and auxiliary information associated with the search request is generated by analyzing the attribute characteristics of the search request. Further, based on the search request and the auxiliary information, the search results corresponding to the search request are obtained. In the above process, the auxiliary information is at least one target attribute feature of the preset industry knowledge corresponding to the search request. Therefore, the search processing method provided by this application mines the attribute features of the search request and combines the attributes corresponding to the preset industry knowledge. Features are introduced into search processing to achieve the purpose of search processing for preset industry knowledge, thereby achieving the technical effect of improving the accuracy of searching for specific industry knowledge, thereby solving the difficulty in injecting industry knowledge into the search processing process in related technologies. Technical issues resulting in less accurate searches.

It is easy to notice that the above general description and the following detailed description are only for illustrating and explaining the present application, and do not constitute a limitation of the present application.

Description of drawings

The drawings described here are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation of the present application. In the attached picture:

Figure 1 is a schematic diagram of the application scenario of the search processing method in this application;

Figure 2 is a flow chart of a search processing method according to Embodiment 1 of the present application;

Figure 3 is a schematic diagram of an optional search processing process according to an embodiment of the present application;

Figure 4 is a flow chart of another search processing method according to Embodiment 2 of the present application;

Figure 5 is a flow chart of another search processing method according to Embodiment 3 of the present application;

Figure 6 is a schematic structural diagram of a search processing device according to Embodiment 4 of the present application;

Figure 7 is a schematic structural diagram of an optional search processing device according to Embodiment 4 of the present application;

Figure 8 is a schematic structural diagram of another search processing device according to Embodiment 4 of the present application;

Figure 9 is a schematic structural diagram of yet another search processing device according to Embodiment 4 of the present application;

Figure 10 is a structural block diagram of a computer terminal according to Embodiment 5 of the present application.

Detailed ways

In order to enable those in the technical field to better understand the solutions of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only These are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts should fall within the scope of protection of this application.

It should be noted that the terms "first", "second", etc. in the description and claims of this application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It is to be understood that the data so used are interchangeable under appropriate circumstances so that the embodiments of the application described herein can be practiced in sequences other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, e.g., a process, method, system, product, or apparatus that encompasses a series of steps or units and need not be limited to those explicitly listed. Those steps or elements may instead include other steps or elements not expressly listed or inherent to the process, method, product or apparatus.

The technical solution provided by this application is mainly implemented using large model technology. The large model here refers to a deep learning model with large-scale model parameters, which can usually include hundreds of millions, tens of billions, hundreds of billions, trillions or even hundreds of thousands. More than 100 million model parameters. The large model can also be called the cornerstone model/foundation model. It is pre-trained through large-scale unlabeled corpus to produce a pre-trained model with more than 100 million parameters. This model can adapt to a wide range of downstream tasks. , the model has good generalization ability, such as large-scale language model (Large Language Model, LLM), multi-modal pre-training model (multi-modal pre-training model), etc.

It should be noted that when a large model is actually used, the pre-trained model can be fine-tuned with a small number of samples, so that the large model can be applied to different tasks. For example, large models can be widely used in fields such as Natural Language Processing (NLP) and computer vision. Specifically, they can be used in areas such as Visual Question Answering (VQA), Image Caption (IC), and image generation. It can also be widely used in natural language processing tasks such as text-based sentiment classification, text summary generation, and machine translation. Therefore, the main application scenarios of large models include but are not limited to digital assistants, intelligent robots, search, online education, office software, e-commerce, intelligent design, etc.

First, some nouns or terms that appear in the description of the embodiments of this application are applicable to the following explanations.

Inference process: The process in which the model produces corresponding output through data input in actual scenarios.

Large language model (LLM): refers to an artificial intelligence model with a certain degree of general capabilities.

Differential Search Index (DSI): refers to a search system that uses large models to replace traditional search engines.

Named Entity Recognition (NER): refers to a technology used to extract named entities from text.

Text fragment (prompt): used as the input initial text of the large model, used to guide the large model to generate specific content.

Feature: refers to the attribute information obtained after attribute marking of the query.

Example 1

According to an embodiment of the present application, a search processing method is provided. It should be noted that the steps shown in the flow chart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although the steps in the flow chart A logical order is shown, but in some cases the steps shown or described may be performed in a different order than herein.

Considering that the large model has a large amount of model parameters and the computing resources of the mobile terminal are limited, the above search processing method provided by the embodiment of the present application can be applied to the application scenario as shown in Figure 1, but is not limited to this. In the application scenario shown in Figure 1, the large model is deployed in the server 10. The server 10 can connect one or more client devices 20 through a local area network connection, a wide area network connection, an Internet connection, or other types of data networks. This The client device 20 may include but is not limited to: smart phones, tablet computers, notebook computers, PDAs, personal computers, smart home devices, vehicle-mounted devices, etc. The client device 20 can interact with the user through a graphical user interface to call the large model, thereby implementing the method provided by the embodiment of the present application.

Under the above operating environment, this application provides a search processing method as shown in Figure 2. Figure 2 is a flow chart of a search processing method according to Embodiment 1 of the present application. As shown in Figure 2, the search processing method includes:

Step S21, obtain the search request;

Step S22, perform attribute feature analysis on the search request, and generate auxiliary information associated with the search request, where the auxiliary information is at least one target attribute feature of the preset industry knowledge corresponding to the search request;

Step S23: Obtain the search results corresponding to the search request based on the search request and the auxiliary information.

The above search request may be a query request in the target application scenario. The above target application scenarios can be, but are not limited to: scenarios involving LLM-based search in fields such as e-commerce, education, medical care, conferences, social networks, financial products, logistics and navigation. Correspondingly, the preset industry knowledge is the industry knowledge corresponding to the above target application scenario.

The above auxiliary information is obtained from preset industry knowledge and at least one target attribute feature. The at least one target attribute feature includes attribute features in multiple dimensions obtained by performing attribute feature mining on the search request. The combination of the above search request and the auxiliary information serves as the input of the search model (which may be LLM), and the above search results are obtained from the output of the search model. It can be seen that the above-mentioned search processing process for the search request takes into account the preset industry knowledge and the multiple attribute characteristics corresponding to the search request. Therefore, the above-mentioned search processing process can have good intelligent search performance in the application scenario of the preset industry. That is to say, accurate search results can be obtained to meet the personalized search needs of specific industries.

In the embodiment of the present application, the search request is first obtained, and auxiliary information associated with the search request is generated by analyzing the attribute characteristics of the search request. Further, based on the search request and the auxiliary information, the search results corresponding to the search request are obtained. In the above process, the auxiliary information is at least one target attribute feature of the preset industry knowledge corresponding to the search request. Therefore, the search processing method provided by this application mines the attribute features of the search request and combines the attributes corresponding to the preset industry knowledge. Features are introduced into search processing to achieve the purpose of search processing for preset industry knowledge, thereby achieving the technical effect of improving the accuracy of searching for specific industry knowledge, thereby solving the difficulty in injecting industry knowledge into the search processing process in related technologies. Technical issues resulting in less accurate searches.

In the embodiment of this application, a system composed of a client device and a server can perform the following steps: the client device sends a search request to the server, the server performs steps corresponding to the search processing method, analyzes the attribute characteristics of the search request, and generates a search request. Associated auxiliary information, where the auxiliary information is at least one target attribute feature of the preset industry knowledge corresponding to the search request, based on the search request and the auxiliary information, obtain the search results corresponding to the search request, and return the search results to the client (or Provide the client with an interface for search processing functions).

It should be noted that, if the operating resources of the client device can meet the deployment and operating conditions of the large model, the embodiments of the present application can be performed on the client device.

In a specific implementation of the above target application scenario, according to the search processing method provided by the embodiment of the present application, the industry knowledge corresponding to the target application scenario is injected into the intermediate reasoning process of DSI to improve the accuracy of the search results and improve the search results. Relevance to corresponding industries. In the above specific implementation method, the search processing process can be shown in Figure 3, including five stages: industry knowledge attribute extraction stage, training small model stage, feature selection stage, thinking chain training stage and online reasoning stage. The search processing method provided by the embodiment of the present application will be further described below by taking this specific implementation as an example.

In an optional embodiment, in step S22, perform attribute feature analysis on the search request and generate auxiliary information associated with the search request, including the following method steps:

Step S221: Extract industry knowledge attributes from the search request to obtain attribute extraction results;

Step S222: Perform attribute annotation on the attribute extraction results to obtain an attribute annotation result, where the attribute annotation result is used to indicate the knowledge attributes of the preset industry knowledge in multiple dimensions;

Step S223: Perform feature selection on the attribute annotation results to obtain auxiliary information.

The knowledge attributes corresponding to the above attribute annotation results at least include: geographical attributes, brand attributes, product attributes, etc. of the preset industry knowledge. In the above optional embodiment, in the industry knowledge attribute extraction stage in the search process, the industry knowledge attributes are extracted from the search request, and the multiple dimensions corresponding to the industry knowledge attributes are determined by the scenario requirements. In the training small model stage, attribute labeling is performed on the extracted multi-dimensional industry knowledge attributes to obtain attribute labeling results. The attribute labeling results can contain multiple groups of labeling data, and each group of labeling data includes industry knowledge attributes and their corresponding labels. In the feature selection stage, feature selection is performed on the attribute annotation results. The obtained auxiliary information is the target annotation data matching the search request in multiple sets of annotation data. During the search processing, the target annotation data is considered to be suitable to help obtain the search results. Industry knowledge.

In an optional embodiment, in step S221, performing industry knowledge attribute extraction on the search request, and obtaining the attribute extraction result includes at least one of the following steps:

Step S2211: Based on the preset industry knowledge, perform word segmentation processing on the search request to obtain the first attribute extraction result;

Step S2212: Based on the preset industry knowledge, perform named entity recognition on the search request to obtain the second attribute extraction result;

Step S2213: Based on the preset industry knowledge, perform synonym expansion on the first attribute extraction result to obtain the third attribute extraction result;

Step S2214: Extract keywords from the search request based on the preset industry knowledge to obtain the fourth attribute extraction result;

Step S2215: Based on the preset industry knowledge, the expression form of the search request is rewritten to obtain the fifth attribute extraction result.

As an exemplary embodiment, the following multiple types of information are extracted from preset industry knowledge (such as e-commerce knowledge, computer technology knowledge, etc.) that help to realize the link between the search request (such as multiple queries) and the target answer text. Attributes.

The first type is word segmentation information. Divide the search request into multiple words according to semantics to obtain word segmentation information. For example, for "on the Nanjing Yangtze River Bridge" contained in the query statement, the word segmentation information obtained after semantic division is: ["on", "Nanjing", "city", "Yangtze River Bridge", "on"].

The second type is Named Entity Recognition (NER) information. Identify the industry knowledge entities included in the search request and obtain NER information. For example, for "Apple mobile phone case" included in the query statement, the NER information obtained by industry knowledge entity recognition includes: ["Apple":"brand", "mobile phone case":"product"].

The third type is synonym information. Perform synonym expansion processing on the word segmentation information to obtain synonym information corresponding to the search request, so as to increase the coverage and accuracy of the search.

The fourth type is keyword information. Identify keywords in search requests (such as query statements) to facilitate subsequent use of the keywords.

The fifth type is to rewrite the information. Rewrite the expression form of the search request to obtain rewritten information, which includes multiple expression sentences corresponding to the same search request.

Through the above-mentioned step S221, step S2211 to step S2215, the embodiment of the present application realizes the extraction of industry knowledge attributes of the search request in multiple dimensions, improves the machine understanding of the search request during the search processing process, and thereby helps to improve the search process. Accuracy and relevance of results.

In an optional embodiment, in step S222, performing attribute annotation on the attribute extraction result to obtain an attribute annotation result includes the following method steps:

Step S2221: Use the annotation model corresponding to the preset industry knowledge to perform attribute annotation on the attribute extraction results to obtain an attribute annotation result, where the attribute annotation result includes: multiple attribute feature combinations.

As an exemplary embodiment, the annotation model corresponding to the above-mentioned preset industry knowledge may be an industry-customized pre-trained small model, for example, a word segmentation model, a NER model for a specified industry, etc. In the training small model stage in the search process, the above-mentioned pre-trained small model is used to perform attribute marking on the attribute extraction results, and multiple attribute feature combinations are obtained. For example, the word segmentation model of a specified industry is used to attribute attribute mark the word segmentation results of search requests corresponding to the specified industry, and n sets of features {f1, f2,...,fn} corresponding to n words in the word segmentation results are obtained, where, Each set of features can contain at least one feature. For example, a set of fi corresponding to the i-th word contains multiple features.

In an optional embodiment, in step S223, feature selection is performed on the attribute annotation results to obtain auxiliary information, including the following method steps:

Step S2231, select multiple candidate feature combinations from the attribute annotation results;

Step S2232, calculate the information gain of multiple candidate feature combinations for the preset result;

Step S2233: Based on the information gain, select at least one target attribute feature from multiple candidate feature combinations to obtain auxiliary information.

As an exemplary embodiment, in the feature selection stage of the search process, for the search request (such as multiple queries), multiple candidate feature combinations are selected from the corresponding attribute annotation results, for example, m features are randomly sampled to obtain combination. Based on m feature combinations, the information gain of each feature combination on the preset result is calculated, where the preset result is the expected answer corresponding to the search request. When the search request contains multiple queries, the above auxiliary information includes at least one target attribute feature F corresponding to each query.

In an optional embodiment, in step S2232, calculating the information gain of multiple candidate feature combinations for the preset result includes the following method steps:

Step S321, obtain the first probability of predicting the preset result based on the search request and the auxiliary information, and obtain the second probability of predicting the preset result based on the search request;

Step S322: Calculate the information gain using the first probability and the second probability.

As an exemplary embodiment, when calculating the information gain of m feature combinations on the preset result, the m feature combinations are recorded as F1 to Fm, and for any one of the feature combinations Fi, the first probability P(doc| query, Fi) and the second probability P(doc|query). The first probability is the probability of obtaining the preset result in response to the search request with auxiliary information. The second probability is the probability of responding to the search without auxiliary information. Probability of requesting a preset result. For any feature combination Fi, the difference between the first probability and the second probability is calculated as the corresponding information gain Hi, that is, Hi=P(doc|query, Fi)-P(doc|query).

In an optional embodiment, in step S2233, selecting at least one target attribute feature from multiple candidate feature combinations based on information gain includes the following method steps:

Step S331: Select the candidate feature combination with the largest information gain from multiple candidate feature combinations to obtain at least one target attribute feature.

As an exemplary embodiment, based on the information gains H1 to Hm corresponding to m feature combinations, the feature combination corresponding to the maximum information gain is selected as at least one target attribute feature F. For example, use the maxarg() function and select the feature combination corresponding to maxarg(Hi) as F.

In an optional embodiment, in step S23, obtaining the search results corresponding to the search request based on the search request and the auxiliary information includes the following method steps:

Step S231: Use a differential search index model to perform industry knowledge inference on the search request and auxiliary information to obtain the target document identifier. The differential search index model is obtained through machine learning training using multiple sets of data. The multiple sets of data are the first training data and the second training data. Mixed data of two training data, the first training data includes: sample prompts and sample document identifiers, the second training data includes: sample prompts, at least one target attribute feature corresponding to the sample prompts and the sample document identifier;

Step S232: Obtain the search result corresponding to the search request based on the target document identifier.

As an exemplary embodiment, the training set used to train the Differential Search Index (DSI) model includes first training data (prompt, target) and second training data (query, F_target), where F_target is represented by query corresponding to At least one target attribute feature F is obtained by splicing the sample document identifier target.

In the thinking chain training stage of search processing, at least one target attribute feature F is used as a thinking chain to train the DSI model. In this way, the DSI model will learn the intermediate reasoning process of industry knowledge, making the search results more industry relevant.

In the online reasoning stage of search processing, when a user searches through the trained DSI model, first, the search request (query_user) input by the user is used as a sample prompt, and the DSI model is used to analyze query_user using a combination of industry knowledge features. Obtain at least one target attribute feature F_user corresponding to the query_user; then, use the splicing result of query_user and at least one target attribute feature F_user as a sample prompt, and obtain the document number corresponding to the query_user and the document content corresponding to the document number through the DSI model As the search result corresponding to query_user. This completes a search query process for query_user.

It is easy to notice that through the above search processing method provided by the embodiment of the present application, preset industry knowledge is injected into the intermediate reasoning process of the DSI model, which improves the accuracy and relevance of the DSI model search results to the specified industry.

In an optional embodiment, a graphical user interface is provided through the terminal device, and the content displayed by the graphical user interface at least partially includes an e-commerce service search dialog box. The search processing method also includes the following method steps:

Step S241, respond to the input operation performed on the e-commerce service search dialog box and determine the e-commerce service search request;

Step S242, in response to the sending operation performed on the e-commerce service search dialog box, perform e-commerce service attribute feature analysis on the e-commerce service search request to generate e-commerce service auxiliary information, and based on the e-commerce service search request and e-commerce service auxiliary information Get e-commerce service search results;

Step S243: Display the e-commerce service search results in the e-commerce service search dialog box.

In the above optional embodiment, the above-mentioned e-commerce service search dialog box can be used to implement an exclusive e-commerce service search function corresponding to the preset industry, such as an intelligent question and answer function set in the application scenario of the preset industry. The intelligent question and answer function Not limited to text Q&A, the above embodiments can also be combined with speech-to-text, text-to-speech, video-to-text, text-to-video and other technologies to implement voice Q&A function, video Q&A function or virtual reality/augmented reality Q&A function.

When an input operation on the e-commerce service search dialog box is detected, an e-commerce service search request is determined based on the input content in the e-commerce service search dialog box. The e-commerce service search request may include at least one query statement.

When a send operation on the e-commerce service search dialog box is detected, a search event is triggered, the e-commerce service attribute characteristics are analyzed for the e-commerce service search request to generate e-commerce service auxiliary information, and the e-commerce service search request is compared with the e-commerce service based on the e-commerce service search request. The service auxiliary information obtains the e-commerce service search results and automatically triggers an event that displays the e-commerce service search results in the e-commerce service search dialog box. The form of displaying the e-commerce service search results to the user can be: displayed in the graphical user interface in the form of text or pictures, converted into audio and output through the audio output device, converted into video and displayed in the graphical user interface.

It should be noted that both the above-mentioned input operation and the above-mentioned sending operation may be operations in which the user touches the display screen of the above-mentioned terminal device with his finger and touches the terminal device. The touch operation may include single-touch, multi-touch, where the touch operation of each touch point may include click, long press, re-press, swipe, etc. The above-mentioned input operation and the above-mentioned sending operation may also be touch operations implemented through input devices such as mouse and keyboard.

Through the above steps S241 to S243, the search processing method provided by the embodiment of the present application can be run on the client, embodying a visual interaction method to realize the search processing function. This visual interaction method can more intuitively combine the preset industries. The question and answer scenario is provided to users through the client, so that users can more easily make e-commerce service search requests and trigger e-commerce service search result queries. This method is user-friendly, convenient and has a good user experience.

It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all It is information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data need to comply with the relevant laws, regulations and standards of relevant countries and regions, and corresponding operation portals are provided for users to choose to authorize or reject.

It should be noted that for the sake of simple description, the foregoing method embodiments are expressed as a series of action combinations. However, those skilled in the art should know that the present application is not limited by the described action sequence. Because in accordance with this application, certain steps may be performed in other orders or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily necessary for this application.

Through the description of the above embodiments, those skilled in the art can clearly understand that the method according to the above embodiments can be implemented by means of software plus a necessary general hardware platform, and of course can also be implemented by hardware. Based on this understanding, the technical solution of the present application can be embodied in the form of a software product in essence or that contributes to the existing technology. The computer software product is stored in a storage medium (such as a read-only memory (Read-Only Memory) , ROM), random access memory (Random Access Memory, RAM), magnetic disk, optical disk), including several instructions to cause a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute this application Methods described in various embodiments.

Example 2

In the operating environment as in Embodiment 1, this application provides another search processing method as shown in Figure 4. Figure 4 is a flow chart of another search processing method according to Embodiment 2 of the present application. As shown in Figure 4, the search processing method includes:

Step S41, obtain the search request;

Step S42: Use the differential search index model to analyze the attribute characteristics of the search request to generate auxiliary information associated with the search request, and perform industry knowledge reasoning on the search request and the auxiliary information to output the target document identification, where the auxiliary information is the search request corresponding Preset at least one target attribute characteristic of industry knowledge;

Step S43: Obtain search results corresponding to the search request based on the target document identifier.

The above search request may be a query request in the target application scenario. The above target application scenarios can be, but are not limited to: scenarios involving LLM-based search in fields such as e-commerce, education, medical care, conferences, social networks, financial products, logistics and navigation. Correspondingly, the preset industry knowledge is the industry knowledge corresponding to the above target application scenario.

The above auxiliary information is obtained from preset industry knowledge and at least one target attribute feature. The target attribute feature includes multiple attribute features obtained by performing attribute feature mining on the search request. The combination of the above search request and auxiliary information serves as the input of the Differential Search Index (DSI) model. The DSI model performs industry knowledge reasoning on the search request and auxiliary information to output the target document identifier, and uses the target document identifier and the corresponding document content as Search results corresponding to the above search request.

It can be seen that the above-mentioned search processing process for the search request takes into account the preset industry knowledge and the multiple attribute characteristics corresponding to the search request. Therefore, the above-mentioned search processing process can have good intelligent search performance in the application scenario of the preset industry. That is to say, accurate search results can be obtained to meet the personalized search needs of specific industries.

In the embodiment of this application, the search request is first obtained, the attribute characteristics of the search request are analyzed by using a differential search index model to generate auxiliary information associated with the search request, and industry knowledge reasoning is performed on the search request and the auxiliary information to output the target document identification , wherein the auxiliary information is at least one target attribute feature of the preset industry knowledge corresponding to the search request, and the search results corresponding to the search request are further obtained based on the target document identification. In the above process, the auxiliary information is at least one target attribute feature of the preset industry knowledge corresponding to the search request. Therefore, the search processing method provided by this application mines the attribute features of the search request and combines the attributes corresponding to the preset industry knowledge. Features are introduced into search processing to achieve the purpose of search processing for preset industry knowledge, thereby achieving the technical effect of improving the accuracy of searching for specific industry knowledge, thereby solving the difficulty in injecting industry knowledge into the search processing process in related technologies. Technical issues resulting in less accurate searches.

In this embodiment of the present application, a system composed of a client device and a server can perform the following steps: the client device sends a search request to the server, the server performs steps corresponding to the search processing method, and uses a differential search index model to attribute attributes to the search request. Analyze to generate auxiliary information associated with the search request, and perform industry knowledge reasoning on the search request and the auxiliary information to output a target document identification, where the auxiliary information is at least one target attribute feature of the preset industry knowledge corresponding to the search request, based on the target document The identifier obtains the search results corresponding to the search request and returns the search results to the client (or provides the client with an interface for the search processing function).

It should be noted that, if the operating resources of the client device can meet the deployment and operating conditions of the large model, the embodiments of the present application can be performed on the client device.

In an optional embodiment, in step S42, a differential search index model is used to perform industry knowledge reasoning on the search request and auxiliary information to output the target document identification, including the following method steps:

Step S421, determine the initial prompt content through the search request, where the initial prompt content is configured based on a preset prompt template;

Step S422, splice the initial prompt content and auxiliary information to obtain the target prompt content;

Step S423: Use the differential search index model to perform industry knowledge inference on the target prompt template to obtain the target document identification.

In the application scenario, the initial prompt content corresponding to the search request (query) is determined through the obtained search request (the initial prompt content can be a text fragment prompt), and the initial prompt content is spliced with the above auxiliary information to obtain Target prompt content and auxiliary information are obtained from at least one target attribute feature (marked as F) of the preset industry knowledge corresponding to the search request. Therefore, the target prompt content can be expressed as query_F, using the trained DSI model, based on the preset industry Knowledge performs industry knowledge reasoning on the target prompt content query_F, obtains the target document identifier, and then obtains the search results corresponding to the search request (query).

It is easy to notice that through the above search processing method provided by the embodiment of the present application, preset industry knowledge is injected into the intermediate reasoning process of the DSI model, which improves the accuracy and relevance of the DSI model search results to the specified industry.

It should be noted that for the preferred implementation of this embodiment, reference can be made to the relevant description in Embodiment 1, which will not be described again here.

Example 3

In the operating environment as in Embodiment 1, this application provides another search processing method as shown in Figure 5. Figure 5 is a flow chart of another search processing method according to Embodiment 3 of the present application. As shown in Figure 5, the search processing method includes:

Step S51, obtain an e-commerce service search request;

Step S52: Perform e-commerce attribute feature analysis on the e-commerce service search request, and generate e-commerce service auxiliary information associated with the e-commerce service search request, where the e-commerce service auxiliary information is the e-commerce service industry knowledge corresponding to the e-commerce service search request. At least one target e-commerce attribute characteristic;

Step S53: Obtain the e-commerce service search results corresponding to the e-commerce service search request based on the e-commerce service search request and the e-commerce service auxiliary information.

The above-mentioned e-commerce service search request may be a query request in an e-commerce service application scenario. Correspondingly, the e-commerce service industry knowledge is the industry knowledge corresponding to the above-mentioned e-commerce service application scenarios, such as dialogue question and answer knowledge, user review knowledge, etc.

The above-mentioned e-commerce service auxiliary information is obtained from e-commerce service industry knowledge and at least one target e-commerce attribute feature. The target attribute feature includes multiple attribute features obtained by performing attribute feature mining on the e-commerce service search request. The combination of the above-mentioned e-commerce service search request and the e-commerce service auxiliary information is used as the input of the search model (which may be LLM), and the above-mentioned e-commerce service search results are obtained from the output of the search model. It can be seen that the above search processing process for the e-commerce service search request takes into account the e-commerce service industry knowledge and the multiple attribute characteristics corresponding to the e-commerce service search request. Therefore, the above search processing process can be used within the application scenario of the preset industry. It has good intelligent search performance, that is, it can obtain accurate e-commerce service search results and meet the personalized search needs of the e-commerce service industry.

In the embodiment of this application, the e-commerce service search request is first obtained, and the e-commerce service search request is analyzed by the e-commerce attribute characteristics to generate e-commerce service auxiliary information associated with the e-commerce service search request, and further based on the e-commerce service search request and e-commerce service auxiliary information to obtain e-commerce service search results corresponding to the e-commerce service search request. In the above process, the e-commerce service auxiliary information is at least one target e-commerce attribute feature of the e-commerce service industry knowledge corresponding to the e-commerce service search request. Therefore, the search processing method provided by this application mines the attributes of the e-commerce service search request. Features, attribute features corresponding to e-commerce service industry knowledge are introduced into search processing, achieving the purpose of search processing for e-commerce service industry knowledge, thereby realizing a technology that improves the accuracy of searching for e-commerce service industry knowledge effect, thereby solving the technical problem in related technologies that it is difficult to inject industry knowledge into the search processing process, resulting in low search accuracy.

In the embodiment of this application, the system composed of a client device and a server can perform the following steps: the client device sends an e-commerce service search request to the server, the server executes the steps corresponding to the search processing method, and electronically processes the e-commerce service search request. Analysis of business attribute characteristics to generate e-commerce service auxiliary information associated with the e-commerce service search request, where the e-commerce service auxiliary information is at least one target e-commerce attribute feature of the e-commerce service industry knowledge corresponding to the e-commerce service search request, based on the e-commerce service search request The e-commerce service search request and the e-commerce service auxiliary information are obtained, the e-commerce service search results corresponding to the e-commerce service search request are obtained, and the e-commerce service search results are returned to the client (or a user interface for the search processing function is provided for the client).

It should be noted that, if the operating resources of the client device can meet the deployment and operating conditions of the large model, the embodiments of the present application can be performed on the client device.

In addition, in addition to e-commerce service scenarios, the search processing method provided in Embodiment 1 can also be extended to legal question and answer scenarios, medical service scenarios, educational question and answer scenarios, etc.

In the legal question and answer scenario, the search processing method includes: obtaining a legal question and answer search request; performing an e-commerce attribute feature analysis on the legal question and answer search request, and generating legal question and answer auxiliary information associated with the legal question and answer search request, where the legal question and answer auxiliary information is legal question and answer At least one target e-commerce attribute characteristic of the legal question and answer industry knowledge corresponding to the search request; based on the legal question and answer search request and the legal question and answer auxiliary information, obtain the legal question and answer search results corresponding to the legal question and answer search request.

In the medical service scenario, the search processing method includes: obtaining a medical service search request; analyzing the e-commerce attribute characteristics of the medical service search request, and generating medical service auxiliary information associated with the medical service search request, where the medical service auxiliary information is medical service At least one target e-commerce attribute characteristic of the medical service industry knowledge corresponding to the search request; and based on the medical service search request and the medical service auxiliary information, obtain the medical service search results corresponding to the medical service search request.

In the educational Q&A scenario, the search processing method includes: obtaining an educational Q&A search request; performing an e-commerce attribute feature analysis on the educational Q&A search request, and generating educational Q&A auxiliary information associated with the educational Q&A search request, where the educational Q&A auxiliary information is educational Q&A At least one target e-commerce attribute characteristic of the education Q&A industry knowledge corresponding to the search request; based on the education Q&A search request and the education Q&A auxiliary information, obtain the education Q&A search results corresponding to the education Q&A search request.

It should be noted that for the preferred implementation of this embodiment, reference can be made to the relevant description in Embodiment 1, which will not be described again here.

Example 4

According to an embodiment of the present application, a device embodiment for implementing the above search processing method is also provided. Figure 6 is a schematic structural diagram of a search processing device according to Embodiment 4 of the present application. As shown in Figure 6, the device includes:

Obtain module 601, used to obtain search requests;

The generation module 602 is configured to perform attribute feature analysis on the search request and generate auxiliary information associated with the search request, where the auxiliary information is at least one target attribute feature of the preset industry knowledge corresponding to the search request;

The processing module 603 is used to obtain search results corresponding to the search request based on the search request and auxiliary information.

Optionally, the above-mentioned generation module 602 is also used to: extract industry knowledge attributes from the search request to obtain attribute extraction results; perform attribute annotation on the attribute extraction results to obtain attribute annotation results, where the attribute annotation results are used to indicate the preset industry. Knowledge attributes of knowledge in multiple dimensions; feature selection is performed on the attribute annotation results to obtain auxiliary information.

Optionally, the above-mentioned generation module 602 is also used to: perform word segmentation processing on the search request based on the preset industry knowledge to obtain the first attribute extraction result; perform named entity recognition on the search request based on the preset industry knowledge to obtain the second attribute Extraction results; based on the preset industry knowledge, perform synonym expansion on the first attribute extraction result to obtain the third attribute extraction result; based on the preset industry knowledge, perform keyword extraction on the search request to obtain the fourth attribute extraction result; based on the preset Industry knowledge is used to rewrite the search request to obtain the fifth attribute extraction result.

Optionally, the above-mentioned generation module 602 is also configured to use an annotation model corresponding to preset industry knowledge to perform attribute annotation on the attribute extraction results to obtain an attribute annotation result, where the attribute annotation result includes: multiple attribute feature combinations.

Optionally, the above-mentioned generation module 602 is also used to: select multiple candidate feature combinations from the attribute annotation results; calculate the information gain of the multiple candidate feature combinations for the preset result; and select from the multiple candidate feature combinations based on the information gain. At least one target attribute feature is used to obtain auxiliary information.

Optionally, the above-mentioned generation module 602 is also configured to: obtain the first probability of predicting the preset result based on the search request and the auxiliary information, and obtain the second probability of predicting the preset result based on the search request; using the first probability and the third probability The information gain is obtained by calculating the second probability.

Optionally, the above-mentioned generation module 602 is also used to select a candidate feature combination with the largest information gain from multiple candidate feature combinations to obtain at least one target attribute feature.

Optionally, the above processing module 603 is also used to: use a differential search index model to perform industry knowledge reasoning on the search request and auxiliary information to obtain the target document identification, wherein the differential search index model is obtained through machine learning training using multiple sets of data. The set of data is a mixture of first training data and second training data. The first training data includes: sample prompts and sample document identifiers. The second training data includes: sample prompts, at least one target attribute feature corresponding to the sample prompts, and the sample document. Identification; obtain the search results corresponding to the search request based on the target document identification.

Optionally, Figure 7 is a schematic structural diagram of an optional search processing device according to Embodiment 4 of the present application. As shown in Figure 7, in addition to all the modules shown in Figure 6, the device also includes: a visualization module 604. A graphical user interface is provided through the terminal device, and the content displayed by the graphical user interface at least partially includes an e-commerce service search dialog box. The above-mentioned visualization module 604 is used to: respond to the input operation performed on the e-commerce service search dialog box, determine the search Request; respond to the send operation performed on the e-commerce service search dialog box, perform attribute feature analysis on the search request to generate auxiliary information, and obtain search results based on the search request and the auxiliary information; display the search results in the e-commerce service search dialog box.

In the embodiment of this application, the search request is first obtained through the acquisition module, the attribute characteristics of the search request are analyzed through the generation module, and the auxiliary information associated with the search request is generated. The processing module is further used to obtain the search request based on the search request and the auxiliary information. Request the corresponding search results. In the above process, the auxiliary information is at least one target attribute feature of the preset industry knowledge corresponding to the search request. Therefore, the search processing method provided by this application mines the attribute features of the search request and combines the attributes corresponding to the preset industry knowledge. Features are introduced into search processing to achieve the purpose of search processing for preset industry knowledge, thereby achieving the technical effect of improving the accuracy of searching for specific industry knowledge, thereby solving the difficulty in injecting industry knowledge into the search processing process in related technologies. Technical issues resulting in less accurate searches.

It should be noted here that the above-mentioned acquisition module 601, generation module 602 and processing module 603 correspond to steps S21 to step S23 in Embodiment 1. The examples and application scenarios implemented by the three modules and the corresponding steps are the same, but they are not It is limited to the content disclosed in the above-mentioned Embodiment 1. It should be noted that the above-mentioned modules or units may be hardware components or software components stored in a memory (for example, the memory 104) and processed by one or more processors (for example, the processors 102a, 102b, ..., 102n) , the above module can also be run in the computer terminal 10 provided in Embodiment 1 as part of the device.

According to the embodiment of the present application, another device embodiment for implementing the search processing method in the above-mentioned Embodiment 2 is also provided. Figure 8 is a schematic structural diagram of another search processing device according to Embodiment 4 of the present application. As shown in Figure 8, the device includes:

Obtain module 801, used to obtain search requests;

The generation module 802 is used to perform attribute feature analysis on the search request using a differential search index model to generate auxiliary information associated with the search request, and perform industry knowledge reasoning on the search request and the auxiliary information to output a target document identifier, where the auxiliary information is the search request Request at least one target attribute characteristic of the corresponding preset industry knowledge;

The processing module 803 is used to obtain the search results corresponding to the search request based on the target document identifier.

Optionally, the above-mentioned generation module 802 is also used to: determine the initial prompt content through a search request, wherein the initial prompt content is configured based on a preset prompt template; splice the initial prompt content and auxiliary information to obtain the target prompt template; use differential The search index model performs industry knowledge reasoning on the target prompt template to obtain the target document identification.

It should be noted here that the above-mentioned acquisition module 801, generation module 802 and processing module 803 correspond to steps S41 to step S43 in Embodiment 2. The examples and application scenarios implemented by the three modules and the corresponding steps are the same, but they are not the same. It is limited to the content disclosed in the above-mentioned Embodiment 2. It should be noted that the above-mentioned modules or units may be hardware components or software components stored in a memory (for example, the memory 104) and processed by one or more processors (for example, the processors 102a, 102b, ..., 102n) , the above module can also be run in the computer terminal 10 provided in Embodiment 1 as part of the device.

According to the embodiment of the present application, another device embodiment for implementing the search processing method in the above-mentioned Embodiment 3 is also provided. Figure 9 is a schematic structural diagram of another search processing device according to Embodiment 4 of the present application. As shown in Figure 9, the device includes:

Acquisition module 901, used to obtain e-commerce service search requests;

The generation module 902 is used to perform e-commerce attribute feature analysis on the e-commerce service search request, and generate e-commerce service auxiliary information associated with the e-commerce service search request, where the e-commerce service auxiliary information is the e-commerce corresponding to the e-commerce service search request. At least one target e-commerce attribute characteristic of service industry knowledge;

The acquisition module 903 is configured to obtain the e-commerce service search results corresponding to the e-commerce service search request based on the e-commerce service search request and the e-commerce service auxiliary information.

It should be noted here that the above-mentioned acquisition module 901, generation module 902 and processing module 903 correspond to steps S51 to step S53 in Embodiment 3. The examples and application scenarios implemented by the three modules and the corresponding steps are the same, but they are not It is limited to the content disclosed in the above-mentioned Embodiment 3. It should be noted that the above-mentioned modules or units may be hardware components or software components stored in a memory (for example, the memory 104) and processed by one or more processors (for example, the processors 102a, 102b, ..., 102n) , the above module can also be run in the computer terminal 10 provided in Embodiment 1 as part of the device.

It should be noted that for the preferred implementation of this embodiment, reference can be made to the relevant descriptions in Embodiment 1, Embodiment 2, or Embodiment 3, which will not be described again here.

Example 5

According to an embodiment of the present application, a computer terminal is also provided, and the computer terminal may be any computer terminal device in a group of computer terminals. Optionally, in this embodiment, the above computer terminal can also be replaced by a terminal device such as a mobile terminal.

Optionally, in this embodiment, the above-mentioned computer terminal may be located in at least one network device among multiple network devices of the computer network.

In this embodiment, the above-mentioned computer terminal can execute the program code of the following steps in the search processing method: obtain the search request; conduct attribute feature analysis on the search request, and generate auxiliary information associated with the search request, where the auxiliary information is the search request corresponding Preset at least one target attribute characteristic of industry knowledge; based on the search request and auxiliary information, obtain the search results corresponding to the search request.

Optionally, FIG. 10 is a structural block diagram of a computer terminal according to Embodiment 5 of the present application. As shown in FIG. 10 , the computer terminal 100 may include: one or more (only one is shown in the figure) processors 1002 , memory 1004, storage controller 1006, and peripheral interface 1008, where the peripheral interface 1008 is connected to the radio frequency module, the audio module and the display.

Among them, the memory 1004 can be used to store software programs and modules, such as the program instructions/modules corresponding to the search processing method and device in the embodiment of the present application. The processor executes various functions by running the software programs and modules stored in the memory. Application and data processing, that is, implementing the above search processing method. Memory 1004 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 1004 may further include memory located remotely relative to the processor, and these remote memories may be connected to the computer terminal 100 through a network. Examples of the above-mentioned networks include but are not limited to the Internet, intranets, local area networks, mobile communication networks and combinations thereof.

The processor 1002 can call the information stored in the memory and the application program through the transmission device to perform the following steps: obtain the search request; perform attribute feature analysis on the search request, and generate auxiliary information associated with the search request, where the auxiliary information is the search request corresponding At least one target attribute feature of the preset industry knowledge; based on the search request and auxiliary information, obtain the search results corresponding to the search request.

Optionally, the above-mentioned processor 1002 can also execute the program code of the following steps: extract industry knowledge attributes from the search request to obtain the attribute extraction results; perform attribute annotation on the attribute extraction results to obtain the attribute annotation results, where the attribute annotation results are expressed in It is used to indicate the knowledge attributes of preset industry knowledge in multiple dimensions; perform feature selection on the attribute annotation results to obtain auxiliary information.

Optionally, the above-mentioned processor 1002 can also execute the program code of the following steps: perform word segmentation processing on the search request based on the preset industry knowledge to obtain the first attribute extraction result; perform named entity recognition on the search request based on the preset industry knowledge. , obtain the second attribute extraction result; based on the preset industry knowledge, perform synonym expansion on the first attribute extraction result, and obtain the third attribute extraction result; based on the preset industry knowledge, perform keyword extraction on the search request, and obtain the fourth attribute extraction Result: Based on the preset industry knowledge, the expression form of the search request is rewritten to obtain the fifth attribute extraction result.

Optionally, the above-mentioned processor 1002 can also execute the program code of the following steps: use an annotation model corresponding to the preset industry knowledge to perform attribute annotation on the attribute extraction results to obtain an attribute annotation result, where the attribute annotation result includes: multiple attributes Feature combination.

Optionally, the above-mentioned processor 1002 can also execute the program code of the following steps: select multiple candidate feature combinations from the attribute annotation results; calculate the information gain of the multiple candidate feature combinations for the preset result; based on the information gain, select from multiple candidate feature combinations. Select at least one target attribute feature from the candidate feature combination to obtain auxiliary information.

Optionally, the above-mentioned processor 1002 can also execute the program code of the following steps: obtaining the first probability of predicting the preset result based on the search request and the auxiliary information, and obtaining the second probability of predicting the preset result based on the search request; using The first probability and the second probability are calculated to obtain the information gain.

Optionally, the above-mentioned processor 1002 can also execute the program code of the following steps: select the candidate feature combination with the largest information gain from multiple candidate feature combinations to obtain at least one target attribute feature.

Optionally, the above-mentioned processor 1002 can also execute the program code of the following steps: using a differential search index model to perform industry knowledge reasoning on the search request and auxiliary information to obtain the target document identification, wherein the differential search index model uses multiple sets of data to pass through the machine Obtained from learning and training, multiple sets of data are mixed data of first training data and second training data. The first training data includes: sample prompts and sample document identifiers. The second training data includes: sample prompts and at least one target corresponding to the sample prompts. Attribute characteristics and sample document identification; obtain search results corresponding to the search request based on the target document identification.

Optionally, the above-mentioned processor 1002 can also execute the program code of the following steps: respond to the input operation performed on the e-commerce service search dialog box, determine the search request; respond to the send operation performed on the e-commerce service search dialog box, respond to the search request Perform attribute feature analysis to generate auxiliary information, and obtain search results based on the search request and auxiliary information; display the search results in the e-commerce service search dialog box.

The processor 1002 can call the information stored in the memory and the application program through the transmission device to perform the following steps: obtain the search request; use the differential search index model to analyze the attribute characteristics of the search request to generate auxiliary information associated with the search request, and perform the search request Request industry knowledge reasoning with the auxiliary information to output a target document identifier, where the auxiliary information is at least one target attribute feature of the preset industry knowledge corresponding to the search request; and obtain the search results corresponding to the search request based on the target document identifier.

Optionally, the above-mentioned processor 1002 can also execute the program code of the following steps: determine the initial prompt content through a search request, wherein the initial prompt content is configured based on a preset prompt template; splice the initial prompt content and auxiliary information to obtain the target Prompt template; use the differential search index model to perform industry knowledge reasoning on the target prompt template to obtain the target document identification.

The processor 1002 can call the information and application programs stored in the memory through the transmission device to perform the following steps: obtain the e-commerce service search request; analyze the e-commerce attribute characteristics of the e-commerce service search request, and generate the e-commerce service search request associated E-commerce service auxiliary information, wherein the e-commerce service auxiliary information is at least one target e-commerce attribute characteristic of the e-commerce service industry knowledge corresponding to the e-commerce service search request; based on the e-commerce service search request and the e-commerce service auxiliary information, the e-commerce service auxiliary information is obtained E-commerce service search results corresponding to the merchant service search request.

Using the embodiments of the present application, a solution for a computer terminal for implementing a search processing method is provided. First, the search request is obtained, and the auxiliary information associated with the search request is generated by analyzing the attribute characteristics of the search request. Further, based on the search request and the auxiliary information, the search results corresponding to the search request are obtained. In the above process, the auxiliary information is at least one target attribute feature of the preset industry knowledge corresponding to the search request. Therefore, the search processing method provided by this application mines the attribute features of the search request and combines the attributes corresponding to the preset industry knowledge. Features are introduced into search processing to achieve the purpose of search processing for preset industry knowledge, thereby achieving the technical effect of improving the accuracy of searching for specific industry knowledge, thereby solving the difficulty in injecting industry knowledge into the search processing process in related technologies. Technical issues resulting in less accurate searches.

Those of ordinary skill in the art can understand that the structure shown in Figure 10 is only illustrative, and the computer terminal can also be a smart phone (such as an Android phone, iOS phone, etc.), a tablet computer, a handheld computer, and a mobile Internet device (Mobile Internet Devices, MID) and other terminal equipment. Figure 10 does not limit the structure of the above computer terminal. For example, the computer terminal 100 may also include more or less components (such as network interfaces, display devices, etc.) than shown in FIG. 10 , or have a different configuration than that shown in FIG. 10 .

Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the hardware related to the terminal device through a program. The program can be stored in a computer-readable storage medium, and the storage medium can Including: flash disk, ROM, RAM, magnetic disk or optical disk, etc.

Example 6

According to an embodiment of the present application, a computer-readable storage medium is also provided. Optionally, in this embodiment, the above-mentioned storage medium can be used to save the program code executed by the search processing method provided in the above-mentioned Embodiment 1, Embodiment 2 or Embodiment 3.

Optionally, in this embodiment, the above storage medium may be located in any computer terminal in a computer terminal group in the computer network, or in any mobile terminal in a mobile terminal group.

Optionally, in this embodiment, the computer-readable storage medium is configured to store program codes for performing the following steps: obtaining a search request; performing attribute feature analysis on the search request, and generating auxiliary information associated with the search request, wherein, The auxiliary information is at least one target attribute feature of the preset industry knowledge corresponding to the search request; based on the search request and the auxiliary information, the search results corresponding to the search request are obtained.

Optionally, in this embodiment, the computer-readable storage medium is configured to store program codes for performing the following steps: extracting industry knowledge attributes from the search request to obtain attribute extraction results; performing attribute annotation on the attribute extraction results, The attribute annotation results are obtained, where the attribute annotation results are used to indicate the knowledge attributes of the preset industry knowledge in multiple dimensions; feature selection is performed on the attribute annotation results to obtain auxiliary information.

Optionally, in this embodiment, the computer-readable storage medium is configured to store program codes for performing the following steps: performing word segmentation processing on the search request based on preset industry knowledge to obtain the first attribute extraction result; Assuming industry knowledge, perform named entity recognition on the search request to obtain the second attribute extraction result; based on the preset industry knowledge, perform synonym expansion on the first attribute extraction result to obtain the third attribute extraction result; based on the preset industry knowledge, perform search Request keyword extraction to obtain the fourth attribute extraction result; based on the preset industry knowledge, rewrite the search request's expression form to obtain the fifth attribute extraction result.

Optionally, in this embodiment, the computer-readable storage medium is configured to store program codes for performing the following steps: using an annotation model corresponding to the preset industry knowledge, performing attribute annotation on the attribute extraction results, and obtaining an attribute annotation result. , where the attribute annotation results include: multiple attribute feature combinations.

Optionally, in this embodiment, the computer-readable storage medium is configured to store program codes for performing the following steps: selecting multiple candidate feature combinations from the attribute annotation results; calculating multiple candidate feature combinations for the preset result Information gain; based on the information gain, select at least one target attribute feature from multiple candidate feature combinations to obtain auxiliary information.

Optionally, in this embodiment, the computer-readable storage medium is configured to store program codes for performing the following steps: obtaining the first probability of predicting the preset result based on the search request and the auxiliary information, and obtaining the first probability based on the search request. Predict the second probability of the preset result; calculate the information gain using the first probability and the second probability.

Optionally, in this embodiment, the computer-readable storage medium is configured to store program codes for performing the following steps: selecting a candidate feature combination with the largest information gain from multiple candidate feature combinations to obtain at least one target attribute feature .

Optionally, in this embodiment, the computer-readable storage medium is configured to store program codes for performing the following steps: using a differential search index model to perform industry knowledge reasoning on the search request and auxiliary information to obtain the target document identification, where , the differential search index model is obtained through machine learning training using multiple sets of data. The multiple sets of data are a mixture of first training data and second training data. The first training data includes: sample prompts and sample document identifiers. The second training data includes : The sample prompt, at least one target attribute feature corresponding to the sample prompt, and the sample document identifier; obtain the search results corresponding to the search request based on the target document identifier.

Optionally, in this embodiment, the computer-readable storage medium is configured to store program codes for performing the following steps: responding to an input operation performed on the e-commerce service search dialog box, determining the search request; responding to the e-commerce service search dialog box The sending operation performed by the search dialog box performs attribute feature analysis on the search request to generate auxiliary information, and obtains search results based on the search request and auxiliary information; the search results are displayed in the e-commerce service search dialog box.

Optionally, in this embodiment, the computer-readable storage medium is configured to store program codes for performing the following steps: obtaining a search request; using a differential search index model to perform attribute feature analysis on the search request to generate a search request-related Auxiliary information, and perform industry knowledge inference on the search request and the auxiliary information to output a target document identifier, where the auxiliary information is at least one target attribute feature of the preset industry knowledge corresponding to the search request; obtain the search corresponding to the search request based on the target document identifier result.

Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for performing the following steps: determining initial prompt content through a search request, wherein the initial prompt content is configured based on a preset prompt template; The initial prompt content and auxiliary information are spliced to obtain the target prompt template; a differential search index model is used to perform industry knowledge reasoning on the target prompt template to obtain the target document identification.

Optionally, in this embodiment, the computer-readable storage medium is configured to store program codes for performing the following steps: obtaining an e-commerce service search request; performing e-commerce attribute feature analysis on the e-commerce service search request, and generating an e-commerce service search request. E-commerce service auxiliary information associated with the e-commerce service search request, where the e-commerce service auxiliary information is at least one target e-commerce attribute feature of the e-commerce service industry knowledge corresponding to the e-commerce service search request; based on the e-commerce service search request and e-commerce Service auxiliary information, obtain the e-commerce service search results corresponding to the e-commerce service search request.

The above serial numbers of the embodiments of the present application are only for description and do not represent the advantages and disadvantages of the embodiments.

In the above-mentioned embodiments of the present application, each embodiment is described with its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components may be combined or may be Integrated into another system, or some features can be ignored, or not implemented. On the other hand, the coupling or direct coupling or communication connection between each other shown or discussed may be through some interfaces, and the indirect coupling or communication connection of the units or modules may be in electrical or other forms.

The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or they may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, each functional unit in each embodiment of the present application can be integrated into one processing unit, each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated units can be implemented in the form of hardware or software functional units.

If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it may be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or contributes to the existing technology, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium , including several instructions to cause a computer device (which can be a personal computer, a server or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of this application. The aforementioned storage media include: U disk, ROM, RAM, mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

The above are only the preferred embodiments of the present application. It should be pointed out that for those of ordinary skill in the art, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications can also be made. should be regarded as the scope of protection of this application.

Claims (14)

1. A search processing method, characterized in that it includes: Get search request; Perform attribute feature analysis on the search request to generate auxiliary information associated with the search request, where the auxiliary information is at least one target attribute feature of the preset industry knowledge corresponding to the search request; Based on the search request and the auxiliary information, a search result corresponding to the search request is obtained. 2. The search processing method according to claim 1, characterized in that performing attribute feature analysis on the search request and generating the auxiliary information associated with the search request includes: Extract industry knowledge attributes from the search request to obtain attribute extraction results; Perform attribute annotation on the attribute extraction results to obtain an attribute annotation result, where the attribute annotation result is used to indicate the knowledge attributes of the preset industry knowledge in multiple dimensions; Feature selection is performed on the attribute annotation results to obtain the auxiliary information. 3. The search processing method according to claim 2, wherein the search request is subjected to industry knowledge attribute extraction, and the attribute extraction result obtained includes one or more of the following: Based on the preset industry knowledge, perform word segmentation processing on the search request to obtain the first attribute extraction result; Based on the preset industry knowledge, perform named entity recognition on the search request to obtain a second attribute extraction result; Based on the preset industry knowledge, perform synonym expansion on the first attribute extraction result to obtain a third attribute extraction result; Based on the preset industry knowledge, perform keyword extraction on the search request to obtain a fourth attribute extraction result; Based on the preset industry knowledge, the expression form of the search request is rewritten to obtain a fifth attribute extraction result. 4. The search processing method according to claim 2, characterized in that, performing attribute annotation on the attribute extraction result, and obtaining the attribute annotation result includes: The annotation model corresponding to the preset industry knowledge is used to perform attribute annotation on the attribute extraction result to obtain the attribute annotation result, where the attribute annotation result includes: a combination of multiple attribute features. 5. The search processing method according to claim 2, characterized in that feature selection is performed on the attribute annotation results to obtain the auxiliary information including: Select multiple candidate feature combinations from the attribute annotation results; Calculate the information gain of the multiple candidate feature combinations for the preset result; Based on the information gain, the at least one target attribute feature is selected from the plurality of candidate feature combinations to obtain the auxiliary information. 6. The search processing method according to claim 5, wherein calculating the information gain of the plurality of candidate feature combinations for the preset result includes: Obtain the first probability of predicting the preset result based on the search request and the auxiliary information, and obtain the second probability of predicting the preset result based on the search request; The information gain is calculated using the first probability and the second probability. 7. The search processing method according to claim 5, characterized in that, based on the information gain, selecting the at least one target attribute feature from the plurality of candidate feature combinations includes: Select the candidate feature combination with the largest information gain from the plurality of candidate feature combinations to obtain the at least one target attribute feature. 8. The search processing method according to claim 1, characterized in that, based on the search request and the auxiliary information, obtaining the search results corresponding to the search request includes: A differential search index model is used to perform industry knowledge inference on the search request and the auxiliary information to obtain the target document identification, wherein the differential search index model is obtained through machine learning training using multiple sets of data, and the multiple sets of data are the first A mixture of training data and second training data. The first training data includes: sample prompts and sample document identifiers. The second training data includes: the sample prompts and at least one target attribute corresponding to the sample prompts. Characteristics and identification of the sample document; Obtain the search result corresponding to the search request based on the target document identifier. 9. The search processing method according to claim 1, characterized in that a graphical user interface is provided through the terminal device, and the content displayed by the graphical user interface at least partially includes an e-commerce service search dialog box, and the search Treatment methods also include: In response to the input operation performed on the e-commerce service search dialog box, determine the e-commerce service search request; In response to the sending operation performed on the e-commerce service search dialog box, perform e-commerce service attribute feature analysis on the e-commerce service search request to generate the e-commerce service auxiliary information, and based on the e-commerce service search request and The e-commerce service auxiliary information obtains the e-commerce service search results; The e-commerce service search results are displayed in the e-commerce service search dialog box. 10. A search processing method, characterized by comprising: Get search request; A differential search index model is used to perform attribute feature analysis on the search request to generate auxiliary information associated with the search request, and industry knowledge reasoning is performed on the search request and the auxiliary information to output a target document identification, wherein: The auxiliary information is at least one target attribute feature of the preset industry knowledge corresponding to the search request; Obtain search results corresponding to the search request based on the target document identifier. 11. The search processing method according to claim 10, wherein using the differential search index model to perform industry knowledge inference on the search request and the auxiliary information to output the target document identification includes: Determine initial prompt content through the search request, wherein the initial prompt content is configured based on a preset prompt template; Splice the initial prompt content and the auxiliary information to obtain the target prompt content; The differential search index model is used to perform industry knowledge reasoning on the target prompt template to obtain the target document identification. 12. A search processing method, characterized by comprising: Get e-commerce service search request; Conduct an e-commerce attribute feature analysis on the e-commerce service search request to generate e-commerce service auxiliary information associated with the e-commerce service search request, where the e-commerce service auxiliary information is the e-commerce service search request corresponding to At least one target e-commerce attribute characteristic of e-commerce service industry knowledge; Based on the e-commerce service search request and the e-commerce service auxiliary information, the e-commerce service search result corresponding to the e-commerce service search request is obtained. 13. An electronic device, characterized in that it includes: Memory, which stores executable programs; A processor, configured to run the program, wherein when the program is run, the search processing method according to any one of claims 1 to 12 is executed. 14. A computer-readable storage medium, characterized in that the computer-readable storage medium includes a stored executable program, wherein when the executable program is running, the device where the computer-readable storage medium is located is controlled to execute the right The search processing method according to any one of claims 1 to 12.