Information search method and apparatus, storage medium, and computer device

Abstract

The present invention relates to the technical field of the Internet. Disclosed are an information search method and apparatus, a storage medium, and a computer device. The method comprises: in response to receiving an information search request, a client sending the information search request to a server; in response to receiving the information search request, the server acquiring search content corresponding to the information search request; on the basis of the search content, generating search guidance and suggestion information by means of a generative large language model, and acquiring a search result corresponding to the information search request, wherein the search guidance and suggestion information comprises at least one of search instruction information, a guide word and a filtering option, and the guide word and the filtering option are interrelated; sending the search guidance and suggestion information and the search result to the client; the client receiving and displaying the search guidance and suggestion information and the search result, wherein the filtering option is used for displaying the search result as a search result corresponding to the content in the filtering option. The method can improve the efficiency and accuracy of information search.

Description

Information search method, device, storage medium and computer equipment Technical Field

[0001] The present invention relates to the field of Internet technology, and in particular to an information search method, device, storage medium and computer equipment. Background Art

[0002] Information search is a key feature of e-commerce platforms. Users can enter arbitrary search terms, such as product names, product categories, merchant names, and so on, into the information search box. The platform's search engine will then provide users with relevant search results for viewing.

[0003] As user search queries become increasingly personalized, many searches often cover very broad information or produce very few matching results. Existing methods for matching information based on keywords in search queries struggle to find search results that are highly relevant to the search query, resulting in weak correlation between search results and the search query, making it difficult to meet user expectations.

[0004] Summary of the Invention

[0005] In view of this, the present application provides an information search method, apparatus, storage medium and computer equipment, the main purpose of which is to solve the technical problem that search content covering very broad information or very few search results is difficult to match accurate search results, resulting in the search results failing to meet the user's search expectations.

[0006] According to a first aspect of the present invention, there is provided an information search method, the method comprising:

[0007] In response to receiving the information search request, the client sends the information search request to the server;

[0008] The server, in response to receiving the information search request, obtains search content corresponding to the information search request;

[0009] The server generates search guidance suggestion information based on the search content using a generative large language model, and obtains search results corresponding to the information search request, wherein the search guidance suggestion information includes at least one of search description information, guide words, and filter items, and the guide words and filter items are associated with each other;

[0010] The server sends the search guidance suggestion information and the search results to the client;

[0011] The client receives and displays the search guidance suggestion information and the search results, and the filter item is used to display the search results as search results corresponding to the content in the filter item.

[0012] According to a second aspect of the present invention, there is provided an information search method, the method comprising:

[0013] In response to receiving the information search request, sending the information search request;

[0014] Receive and display search guidance suggestion information and search results, wherein the search guidance suggestion information is generated based on the search content through a generative large language model, and the search results are obtained through the information search request; the search guidance suggestion information includes at least one of search description information, guide words and filter items; the guide words and the filter items are associated with each other, and the filter items are used to display the search results as search results corresponding to the content in the filter items.

[0015] According to a third aspect of the present invention, there is provided an information search method, the method comprising:

[0016] In response to receiving an information search request, obtaining search content corresponding to the information request;

[0017] generating search guidance suggestion information based on the search content using a generative large language model, and obtaining search results corresponding to the information search request, wherein the search guidance suggestion information includes at least one of search description information, guide words, and filter items, and the guide words and filter items are associated with each other;

[0018] The search guidance suggestion information and the search results are sent to the client, and the filter item is used to display the search results as search results corresponding to the content in the filter item.

[0019] According to a fourth aspect of the present invention, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned information search method when executing the program.

[0020] The present invention provides an information search method, apparatus, storage medium, and computer device. First, in response to an information search request initiated by a user, the information search request is sent to a server so that the server obtains the search content corresponding to the information search request. Then, based on the search content, a generative large language model is used to generate search guidance suggestion information covering various information such as search description information, guide words, and filter items, and obtain search results corresponding to the information search request. Finally, the generated search guidance suggestion information and the obtained search results are sent to a client for display. By combining a generative large language model with an information search scenario, the above method can use the text analysis and text generation capabilities of the generative large language model to decompose and extract information from search content that covers relatively broad information or has very few search results, thereby generating search description information, guide words, and filter items that can guide the user to conduct further information searches. By displaying the search guidance suggestion information, the client can provide users with more accurate and personalized search results and present the search results to users in a more interpretable way, thereby improving the accuracy of the search results and enhancing the user's information search experience. In addition, the above information search method will not interfere with the normal information search link. The user can still view the search results corresponding to the information search request by initiating an information search request.

[0021] The above description is only an overview of the technical solution of this application. In order to more clearly understand the technical means of this application, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of this application more obvious and easy to understand, the following are the technical solutions of this application. Specific implementation method. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0023] FIG1 is a schematic diagram showing a flow chart of an information search method provided by an embodiment of the present invention;

[0024] FIG2 is a schematic diagram showing a scenario of an information search method provided by an embodiment of the present invention;

[0025] FIG3 shows a scenario diagram of an information search method provided by an embodiment of the present invention;

[0026] FIG4 shows a scenario diagram of an information search method provided by an embodiment of the present invention;

[0027] FIG5 is a schematic diagram showing a scenario of an information search method provided by an embodiment of the present invention;

[0028] FIG6 shows a schematic flow chart of an information search method provided by an embodiment of the present invention;

[0029] FIG7 shows a schematic flow chart of an information search method provided by an embodiment of the present invention;

[0030] FIG8 shows a schematic structural diagram of an information search device provided by an embodiment of the present invention;

[0031] FIG9 shows a schematic structural diagram of an information search device provided by an embodiment of the present invention;

[0032] FIG10 shows a schematic structural diagram of a computer device provided by an embodiment of the present invention.

[0033] DETAILED DESCRIPTION

[0034] The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that, unless there is a conflict, the embodiments and features in the embodiments of the present application can be combined with each other.

[0035] In one embodiment, as shown in FIG1 , an information search method is provided, which is described by taking the method applied to a client and a server as an example, including the following steps:

[0036] 101. In response to receiving the information search request, the client sends the information search request to the server.

[0037] Specifically, when the user enters a search content or selects a search content to send through the information search box of the client, the client can detect that the user has initiated an information search request. After detecting that the user has initiated an information search request, the client can send the information search request to the server.

[0038] In this embodiment, users can initiate information search requests in a variety of ways. For example, users can initiate an information search request by selecting a specific search content, by entering and sending a search content, or by clicking a pre-set search entry. Furthermore, the information search request initiated by the user can be a preset artificial intelligence search request or a normal information search request. Regardless of the type of information search request initiated by the user, the information search request will be sent to the server.

[0039] 102. In response to receiving the information search request, the server obtains search content corresponding to the information search request.

[0040] Among them, the search content refers to the content entered or selected by the user when initiating an information search request. The content can be search keywords such as product name, product category, merchant name, etc., or it can be a phrase entered by the user at will, such as "what to eat for a cold" or "what to eat for dinner". Generally speaking, for search content with clear search keywords, the search engine will more accurately recall the corresponding search results for the user. At this time, the correlation between the search results and the search content is high, and the search accuracy is high. In contrast, for phrases entered by the user at will, the search engine is usually unable to accurately match, the correlation between the search results and the search content is low, and the search accuracy is also low. In addition, for some uncommon search keywords, such as "such and such brand milk tea", since this brand of milk tea does not exist within the delivery range of the user's location, the number of search results is also very small or zero (zero or few results). At this time, the search results recalled by the search engine are also less relevant to the search content, and the search accuracy is also low.

[0041] Specifically, in response to receiving the information search request, the server may obtain the search content corresponding to the information search request. The search content may be included in the information search request or sent separately to the server. This embodiment does not specifically limit the method for obtaining the search content.

[0042] 103. The server generates search guidance suggestion information based on the search content through a generative large language model, and obtains search results corresponding to the information search request.

[0043] 104. The server sends the search guidance suggestion information and search results to the client.

[0044] Among them, the search guidance suggestion information includes at least one of the following information: search description information, guide words and filter items. Among them, the search description information refers to the suggested reasons for the search content with a certain word limit, and / or the guidance description information of the extended keywords related to the search content. For example, assuming the search content is "What to eat for a cold", the search description information can be "It is suitable to eat some light food for a cold", "Chicken soup is rich in protein and amino acids, which can improve immunity and anti-inflammatory effects"; assuming the search content is "What to eat for breakfast", the search description information can be "You can eat some nutritious food for breakfast", "Steamed buns and porridge can be an ideal choice for a nutritious breakfast"; assuming the search content is "XX milk tea" and the current search results are few, the search description information can be "The current search results are few, looking for some inspiration for drinking milk tea for you", "As a classic milk tea category, delicious milk tea may be worth a try".

[0045] Furthermore, the guide word refers to an extended keyword generated based on the search content. The extended keyword has a certain correlation with the search content and the search description information, and can be directly used as a search keyword to guide users to conduct further information searches. Generally, it can be a product name, product category, merchant category, etc. For example, assuming the search content is "What to eat for a cold", the search description information is "It is suitable to eat some light foods for a cold", "Chicken soup is rich in protein and amino acids, which can improve immunity and anti-inflammatory effects; green vegetables help to enhance immunity", then the guide word can be "chicken soup, green vegetables" and other lighter product categories or product names. Through the guide word, the user can re-initiate an information search request with a clearer search intention to obtain accurate search results.

[0046] Furthermore, the filter item refers to an extended keyword that is mutually associated with the guide word. The filter item can respond to the user's trigger operation and directly initiate an information search request with the extended keyword corresponding to the filter item as the search keyword, thereby obtaining search results corresponding to the extended keyword. Among them, the filter word can be a product name, product category, merchant category, etc. For example, assuming the search content is "What to eat for a cold", the search description information is "It is suitable to eat some light food for a cold", "Chicken soup is rich in protein and amino acids, which can improve immunity and anti-inflammatory effects; green vegetables help to enhance immunity", and the guide words are "chicken soup, green vegetables, tomatoes" and other product category names, then the filter items can be "old hen soup, boiled green vegetables, tomato scrambled eggs" and other product names. By clicking the filter item "tomato scrambled eggs", the user can directly view the search results corresponding to "tomato scrambled eggs", thereby improving the efficiency of information search.

[0047] Specifically, after obtaining the search content corresponding to the information search request, the server can pre-process the search content and input the search content into the generative large language model in a certain manner, so as to perform keyword expansion, keyword prediction, guidance suggestion generation and other related processing based on the search content through the generative large language model, so as to obtain search guidance suggestion information that can guide the user to conduct further information search, and then send the generated search guidance suggestion information to the client.

[0048] In this embodiment, when inputting search content into the generative large language model, pre-set prompt templates can be used for information input. This allows the generative large language model to better understand the task objectives and confirm the steps for executing the task, thereby outputting search guidance suggestions that are more helpful to the user's information search. In addition, since the search content is input by the user based on their own behavioral habits and is highly random, prompt templates can be designed for different scenarios to improve the accuracy of the information output.

[0049] In addition, after receiving the information search request, the server can also obtain the search results corresponding to the information search request. Among them, obtaining the search results corresponding to the information search request can be directly based on the search content to obtain the search results; it can also be based on part of the information in the search guidance suggestion information generated by the search content to obtain the search results; it can also be based on the search content and / or any intermediate or result information generated by the search content to obtain the search results. It can be understood that, in the absence of conflict and mutual dependence, the order of generating the search guidance suggestion information and obtaining the search results is not restricted; in the case of dependence, they can also be executed one after another. Further, after obtaining the search results corresponding to the information search request, the server can send the search results to the client.

[0050] 105. The client receives and displays the search guidance suggestion information and the search results. The filter item is used to display the search results as search results corresponding to the content in the filter item.

[0051] Specifically, after receiving the search guidance suggestion information and search results sent by the server, the client can display the search guidance suggestion information and search results on a preset page of the client. The preset page can be a human-computer dialogue page, a search results page, or other pages. The search instructions, guide words, filter items, and search results in the search guidance suggestion can be displayed in different locations on the preset page in a pre-set manner. In this embodiment, the filter items in the search guidance suggestion can be used to update the search results displayed on the current page to search results corresponding to the content in the filter items.

[0052] In this embodiment, the client can receive the search guidance suggestion information in real time and dynamically display the search guidance suggestion information received in real time. It can also display all the search guidance suggestion information in full according to the preset layout after receiving the search guidance suggestion information. In addition, the client can receive part of the search guidance suggestion information for display, or receive all the search guidance suggestion information for display. In some scenarios, the search guidance suggestion information displayed by the client may be blocked or hidden due to certain operations of the user. To ensure that the user can see the search guidance suggestion information, the client can display at least part of the search guidance suggestion information. For example, the client can only display the search suggestions and filter items, and hide the guide words, so as to improve the efficiency of the user in obtaining the search guidance suggestion information.

[0053] In one embodiment, step 101 may be implemented in the following manner: in response to detecting that the information search request initiated by the user is a preset artificial intelligence search request, the client sends the information search request to the server.

[0054] Among them, a preset AI search request refers to an information search request that can trigger the server to activate a generative large language model to search and guide the user's search content. Aside from the preset AI search request, all other information search requests are information search requests that execute a normal search link. After an information search request that executes a normal search link is sent to the server, the server can obtain search results corresponding to the information search request. In the process of obtaining search results, relevant AI methods may be used. However, such requests that use AI methods to obtain search results are not preset AI search requests referred to in this embodiment, but rather other types of AI search requests.

[0055] Specifically, when a user enters a search term or selects a search term to send through the client's information search box, the client can detect that the user has initiated an information search request. After detecting that the user has initiated an information search request, the client can determine whether the information search request initiated by the user is a preset artificial intelligence search request. If it is detected that the information search request initiated by the user is a preset artificial intelligence search request, the client can send the information search request to the server.

[0056] In this embodiment, it is possible to detect whether the information search request initiated by the user is a preset artificial intelligence search request in a variety of ways, wherein the preset artificial intelligence search request can be divided into two initiation methods: active and passive. For example, the client can actively initiate a preset artificial intelligence search request by triggering the artificial intelligence search entrance and / or artificial intelligence term displayed on the client, or it can passively determine the information search request initiated by the user as a preset artificial intelligence search request when it detects that the search content input by the user meets the initiation conditions of the preset artificial intelligence search request. Accordingly, the client can also detect whether the information search request initiated by the user is a preset artificial intelligence search request in a variety of ways. For example, the client can identify it by the initiation path of the information search request, the artificial intelligence identifier carried in the information search request, etc., which are not specifically limited here.

[0057] This embodiment can improve the speed of information search request diversion by sending the information search request to the server when it is detected that the information search request initiated by the user is a preset artificial intelligence search request. Specifically, when it is detected that the information search request is not a preset artificial intelligence search request, the information search request can be diverted to the search engine in the server to obtain search results corresponding to the information search request; when it is detected that the information search request is a preset artificial intelligence search request, the information search request is sent to the search engine in the server and the interface of the generative large language model respectively, to obtain search results corresponding to the information search request and search guidance suggestion information generated based on the search content.

[0058] In one embodiment, the method of detecting whether the information search request initiated by the user is a preset artificial intelligence search request in step 101 can be implemented in at least one of the following ways:

[0059] 1011. The client determines that the information search request initiated by the user is a preset artificial intelligence search request in response to the triggering operation of the artificial intelligence search portal and / or artificial intelligence term.

[0060] Specifically, the method of step 1011 is a method of actively initiating a preset artificial intelligence search request, wherein the artificial intelligence search entrance can be designed in a fixed style, or can be designed in a style combined with an artificial intelligence term, and the artificial intelligence term can be obtained based on the search content input by the user. Among them, the artificial intelligence term can be obtained by text matching based on the keywords in the search content input by the user, or can be directly generated by a generative large language model based on the search content input by the user. It can be understood that the setting method and setting location of the artificial intelligence search entrance and / or artificial intelligence term can be set according to actual conditions, and this embodiment does not make specific limitations here.

[0061] For example, as shown in the figure on the left side of Figure 3, when the user clicks on the artificial intelligence search portal "View Recommendations", or clicks on the artificial intelligence entry "What foods should I eat when I have a cold?", or clicks on the artificial intelligence search portal of "What foods should I eat when I have a cold? View recommendations" to initiate an information search operation, the client can respond to the triggering operation of the artificial intelligence search portal or the artificial intelligence entry, determine that the information search request initiated by the user is a preset artificial intelligence search request, and then send the information search request to the server.

[0062] 1012. In the artificial intelligence mode, the client determines that the information search request initiated by the user is a preset artificial intelligence search request in response to the sending operation of the search content.

[0063] Specifically, the method of step 1012 is a method of actively initiating a preset artificial intelligence search request, and the user can enter the artificial intelligence mode in an active or passive manner. For example, the user can actively enter the artificial intelligence mode by clicking on the artificial intelligence search entrance and / or artificial intelligence terms, or can passively enter the artificial intelligence mode when the input search content meets the preset conditions. In artificial intelligence mode, the user can enter any keyword or phrase and initiate an information search request to trigger the preset artificial intelligence search request again. It is understandable that the method of entering or exiting the artificial intelligence mode can be set according to the actual situation, and this embodiment does not make specific limitations here.

[0064] 1013. When the search content input by the user meets the preset conditions, in response to the sending operation of the search content, determine that the information search request initiated by the user is a preset artificial intelligence search request.

[0065] Specifically, step 1013 involves passively initiating a preset artificial intelligence search request, where the preset conditions may include at least one of the following: the keywords in the search content match the preset keywords, and the number of search results corresponding to the search content is less than a preset threshold. Under the first condition, if the keywords in the search content match the preset keywords, it can be determined that the keywords in the search content meet the characteristics of a preset broad keyword, such as "cold, breakfast, lunch, dinner, sour and sweet, sweet and spicy," etc. In this case, the correlation between the search results and the search content is low. Therefore, artificial intelligence can be used to guide the search of these search contents to improve the accuracy of the information search. The preset keywords can be pre-stored in an artificial intelligence offline vocabulary to facilitate retrieval and matching. Furthermore, under the second condition, if the number of search results corresponding to the search content is less than a preset threshold, it can be determined that the current search content has a small number of search results and the correlation between the search results and the search content is also low. In this case, artificial intelligence can also be used to guide the search of these search contents to improve the accuracy of the information search. It is understood that the conditional judgment process can be performed on the client or the server. In addition, the preset conditions can also be set to other conditions, which are not specifically limited in this embodiment.

[0066] For example, as shown in Figure 2, after the user enters the search content "XX milk tea", the client detects that there is only one search result for this brand of milk tea, and this number is less than the preset number threshold. After the information search request is sent, the client can determine that the information search request initiated by the user is a preset artificial intelligence search request based on this condition. For another example, after the user enters the search content "What food to eat when you have a cold?", the client detects that the search content matches the preset keywords "cold, food". After the information search request is sent, the client can also determine that the information search request initiated by the user is a preset artificial intelligence search request based on this condition, and then send the information search request to the server.

[0067] The above-described embodiment uses various methods to detect whether a user-initiated information search request is a preset AI search request. This allows users to participate in AI search scenarios in various ways, thereby increasing their awareness and participation in AI search. Furthermore, by initiating preset AI search requests, this embodiment can improve search accuracy, address the low hit rate of traditional search results, and thus enhance the user's search experience.

[0068] In one embodiment, before step 101, the above information search method further includes the following steps:

[0069] 201. The client responds to receiving the pre-search content input by the user and sends the pre-search content to the server.

[0070] Among them, the client can send the pre-search content to the server during the process of the user's real-time input of the pre-search content. For example, when the user inputs the character "感", the client can send the character "感" to the server, and when the user further inputs the character "冒", the client can send "感冒" to the server.

[0071] 202. The server determines whether the keywords in the pre-search content match the preset keywords, and / or determines whether the number of search results corresponding to the pre-search content is less than the preset quantity threshold.

[0072] Among them, the method for the server to make conditional judgments can be referred to step 1013 and will not be elaborated here.

[0073] 203. When the keywords in the pre-search content match the preset keywords, and / or the number of search results corresponding to the search content is less than the preset quantity threshold, the server sends the artificial intelligence search entry and / or artificial intelligence entries to the client.

[0074] Among them, the artificial intelligence entries can be determined according to the keywords in the pre-search content. In this embodiment, the artificial intelligence entries can be obtained by means of text matching based on the pre-search content input by the user, or directly generated by a generative large language model based on the pre-search content input by the user. For example, when the pre-search content input by the user is "感冒", the artificial intelligence entries can be "感冒适合吃什么", or "感冒时吃哪些药", etc.

[0075] In this embodiment, when the server detects that the keywords in the pre-search content match the preset keywords, and / or the number of search results corresponding to the pre-search content is less than the preset quantity threshold, it can also send an artificial intelligence identifier to the client. The client can determine that the current information search request is a preset artificial intelligence search request based on this artificial intelligence identifier, thereby improving the detection speed of the preset artificial intelligence search request.

[0076] 204. The client displays the artificial intelligence search entry and / or artificial intelligence entries.

[0077] Referring to the image on the left side of FIG3 , the client can display an artificial intelligence search portal and / or artificial intelligence terms on the search association page to facilitate users to initiate preset artificial intelligence search requests during the search process. The client can also display the artificial intelligence search portal and / or artificial intelligence terms on other pages. For example, the client can display the artificial intelligence search portal at a fixed location on the homepage to facilitate users to directly initiate preset artificial intelligence search requests on the homepage. It is understood that this embodiment does not specifically limit the display method and display location of the artificial intelligence search portal and / or artificial intelligence terms.

[0078] In one embodiment, the information search method may further include at least one of the following methods:

[0079] 205. The client enters the artificial intelligence mode in response to the triggering operation of the artificial intelligence search portal and / or artificial intelligence term.

[0080] 206. When the search content input by the user meets the preset conditions, the client enters the artificial intelligence mode in response to at least one of the sending operation, the modifying operation and the replacing operation of the search content.

[0081] Specifically, when a user clicks on the AI ​​search entry and / or AI term displayed on the client, or when the search content entered meets preset conditions, the client can control the application to enter AI mode. In AI mode, the user can freely modify and replace the search content. After the user sends the modified or replaced search content, the client can detect that the information search request initiated by the user is a preset AI search request and then send the information search request to the server. It will be understood that in other embodiments, when the user clicks on the AI ​​search entry and / or AI term, or when the search content entered meets the preset conditions, the client can simply determine the information search request initiated by the user as the preset AI search request, without entering AI mode. Then, if the user clicks on the AI ​​search entry and / or AI term again, or if the search content entered meets the preset conditions, the next information search request initiated will be determined to be a preset AI search request.

[0082] In the above embodiment, when the user actively or passively initiates a preset artificial intelligence search request, it means that the user may not have a clear search intention at this time. In this scenario, the user may constantly change his search content until he sees the search results that he is interested in. In response to this scenario, this embodiment directly enters the artificial intelligence mode after detecting that the user has initiated a preset artificial intelligence search request, allowing the user to arbitrarily change or replace his search content in the artificial intelligence mode, and for each search content entered, the client can determine the information search request initiated by the user as a preset artificial intelligence search request, thereby guiding the user to conduct a more specific information search, thereby improving the user's search efficiency and improving the user's search experience.

[0083] In one embodiment, the method for the server to generate search guidance suggestion information using the generative large language model in step 103 may be implemented by the following method:

[0084] 1031. The server inputs the search content, the user information initiating the information search request, and the preset prompt template into the generative large language model, so that the generative large language model generates user behavior preference information based on the prompt template and the user information.

[0085] 1032. The generative large language model running in the server generates search guidance suggestion information based on the prompt template, user behavior preference information and search content.

[0086] Before generating search guidance suggestion information based on the search content, the server may first obtain a prompt template (i.e., a prompt) for generating the search guidance suggestion information, as well as user information of the user initiating the information search request. The user information may be obtained from a database or server log with user authorization. In this embodiment, the user information may include at least one of user behavior information, user profile information, and user basic information.

[0087] Specifically, in the process of generating search guidance suggestion information through a generative large language model (hereinafter referred to as the large model), user behavior preference information can be generated with the help of a preset prompt template, such as generating user preferences for keywords, product names, product categories, and merchant brands. Then, with the help of user preference information combined with the search content input by the user, the search results that the user is interested in are predicted, so as to obtain search guidance suggestion information that is closer to the search content and more relevant to the user's interests.

[0088] The above embodiment generates user behavior preference information with the help of preset prompt templates and user information, and then generates search guidance suggestion information through user behavior preference information and search content. This can make it easier for the large model to understand user behavior preferences, task goals, and the execution steps of task goals, thereby generating search guidance suggestion information that is closer to the search content and closer to user preferences, thereby improving the accuracy of the search guidance suggestion information and its matching degree with the user, and improving the user search experience.

[0089] In one embodiment, the prompt template may include at least one behavior preference dimension and execution steps corresponding to the behavior preference dimension. Step 1031 may be implemented as follows: the server generates user behavior preference information for at least one behavior preference dimension based on user information, using the execution steps corresponding to the behavior preference dimension. The behavior preference dimension of the user behavior preference information may include at least one of keyword preference, product category preference, merchant brand preference, delivery speed preference, order time preference, specific product preference, and product attribute preference.

[0090] Specifically, the prompt template can include at least one behavioral preference dimension and the execution steps corresponding to the behavioral preference dimension, so that the large model can quantitatively analyze user information through multiple behavioral preference dimensions and the execution steps corresponding to each behavioral preference dimension, thereby obtaining user behavioral preference information that matches the search scenario. In addition, the prompt template can also include role information so that the model can more accurately understand the task objectives. For example, the prompt template can include the following content:

[0091] 1. Character Setting

[0092] 1) Role: You are an expert in food ordering, good at giving guidance and suggestions based on different broad intent scenarios, but be careful not to give advice related to medication and diet.

[0093] 2) Personality: You come across as an expert who speaks plainly and doesn’t use jargon.

[0094] 2. Task training

[0095] Based on the user's ordering records on the food delivery platform in the last 7 / 14 / 30 days, summarize the user's ordering preferences during this period in the given format and make corresponding recommendations based on their eating patterns.

[0096] 1) Behavior Summary

[0097] User search preferences: User search engine preferences over the past 7 / 14 / 30 days, including but not limited to the following dimensions:

[0098] 1. Keyword preference: Users may tend to use specific keywords or phrases when expressing their search intent.

[0099] 2. Food and Beverage Category Preference: Collect statistics on the food and beverage categories that users order most frequently, such as Chinese, Western, fast food, and healthy light meals.

[0100] 3. Merchant brand preference: Collect the merchant brands that users order the most from to identify brand loyalty.

[0101] 4. Logistics delivery speed: Calculate the average delivery time of user orders and analyze user sensitivity to delivery speed.

[0102] 5. Meal time regularity: Organize users’ ordering times and analyze their daily eating patterns, such as breakfast, lunch, and dinner time periods.

[0103] 6. Specific dish preferences: List the specific dishes that users frequently order to understand their preferences for specific foods.

[0104] 7. Food taste preferences: Extract the user's preferred taste types from the order options, such as spicy, sweet, light or various special seasonings.

[0105] In the above embodiment, by setting the execution steps corresponding to the behavior preference dimensions in the prompt template, the macro model can be facilitated to accurately generate multiple dimensions of user behavior preference information that match the user information through the execution steps corresponding to each preference dimension. Furthermore, by setting the preference dimensions in advance, it is possible to avoid hallucinations in the macro model or unstable content generated by the macro model, thereby making the user behavior preference information more closely matched to the search scenario. At the same time, it is also convenient to obtain search guidance suggestion information that matches the user's own characteristics based on the user behavior preference information, thereby improving the accuracy and matching degree of information search.

[0106] In one embodiment, the prompt template may further include a task objective and steps for executing the task objective. Step 1032 may be implemented as follows: the server constructs a semantic expansion network based on the task objective and the steps for executing the task objective, keywords in the search content, and related words corresponding to the keywords, and predicts at least one search expansion term and a corresponding guidance suggestion based on the semantic expansion network based on user behavior preference information and / or user location information, and generates search guidance suggestion information based on the search expansion term and the corresponding guidance suggestion.

[0107] Specifically, the prompt template can also include the task goal and the execution steps of the task goal, so that the large model can accurately obtain search guidance suggestion information that matches the task goal based on the execution steps of the task goal and the user's behavior preference information and the search content entered by the user. For example, continuing with the example of the prompt template, the prompt template can also include the following content:

[0108] 2) Goal: Provide keyword guidance suggestions

[0109] We use synonyms, hyponyms, and related words of the user's input phrase to build a semantic expansion network. We also use the context to predict the user's likely search expansion terms. We combine the collected user context information, such as location and search history, to provide guidance for keywords:

[0110] Suggestions for guiding broad searches for catering terms:

[0111] 1) Key phrase: Reason for the suggestion, briefly describe the reason in one sentence

[0112] 2) Guidance words: Provide at least three guidance suggestions based on the user's search keywords.

[0113] In the above embodiment, the related words corresponding to the keywords in the search content may be synonyms, hyponyms, and conjunctions of the keywords. For example, assuming the input content is "what food to eat when you have a cold", the keywords in the search content may include "cold" and "food", which can be expanded to include "sick", "fever", "diet", "dishes" and other words. Through the expanded words, a semantic expansion network can be constructed. On the basis of the semantic expansion network, combined with the user's preferences for food, brands and keywords in the past period of time, and combined with the user's location, search expansion words such as "chicken soup, green vegetables, tomatoes" and guidance suggestions corresponding to the search expansion words can be predicted, thereby obtaining search guidance suggestion information that matches the search content and user behavior preferences.

[0114] The above embodiment constructs a semantic expansion network using related words corresponding to keywords in the search content. Based on this semantic expansion network, the user's past preferences for food, brands, and keywords, as well as the user's location, are combined to predict search expansion terms. This allows the predicted keywords to no longer be limited to the search content itself but to be extended to the entire semantic network, effectively expanding the scope of keyword prediction. Furthermore, the prediction results can be made more closely aligned with the user's behavioral preferences and conform to the geographic characteristics of the user's location, thereby improving the accuracy of keyword prediction.

[0115] In one embodiment, the prompt template also includes a task output format. The method of generating search guidance suggestion information based on the search expansion term and the guidance suggestion corresponding to the search expansion term in step 1032 can be implemented in at least one of the following ways:

[0116] 301. The server generates and outputs search description information, at least one guide word, and at least one filter item according to a task output format and based on the search expansion word and the guide suggestion corresponding to the search expansion word.

[0117] 302. The server generates and outputs search description information and at least one guide word according to the task output format, based on the search expansion word and the guide suggestion corresponding to the search expansion word. Then, based on the guide word, text matching is performed through preset product names and / or product categories and / or merchant categories to obtain at least one filter item.

[0118] Specifically, the server can directly generate the search guidance suggestion information according to the task output format in the prompt template and send it to the client, or it can output the basic content of the search guidance suggestion information, and then package it according to a preset typesetting format and send it to the client, or it can combine the two information output methods, that is, first output the content output by the large language model, and then output the typeset search guidance suggestion information to the client according to a preset typesetting format. This embodiment does not make specific restrictions here.

[0119] In the above embodiment, outputting search guidance suggestion information in a task output format and sending it to the client can increase the speed at which users can view the guidance suggestion information. Matching the guide words with preset product names and / or product categories and / or merchant categories and generating filter items can improve the matching degree between the filter items and the search keywords, thereby improving the relevance between the search results and the search content, and thus improving the efficiency and accuracy of information search. In this embodiment, the server can first output the generated content such as the search description information and guide words, and then output the filter items, without having to wait until all information is generated before sending.

[0120] In one embodiment, the method of generating filter items using guide words in step 302 may be implemented in at least one of the following ways:

[0121] 401. The server uses the product names and / or product categories and / or merchant categories that match the guide words as screening items.

[0122] 402. The server performs a text match between the product name and / or product category and / or merchant category that matches the guide word and the name corresponding to the preset icon, and uses the product name and / or product category and / or merchant category, and the icon that matches the product name and / or product category and / or merchant category as filtering items.

[0123] Specifically, when the server generates filtering items through guide words, it can directly send the product names and / or product categories and / or merchant categories that match the guide words as filtering items to the client for display; it can also send the product names and / or product categories and / or merchant categories that match the guide words, as well as the icons corresponding to the product names and / or product categories and / or merchant categories as filtering items to the client for display.

[0124] For example, as shown in Figure 5, after text matching the guide word "chicken soup" with the preset product name and / or product category name, the filter item "old hen soup" can be obtained. By text matching the filter item "old hen soup" with the name corresponding to the preset icon, the icon corresponding to "old hen soup" is not obtained. At this time, "old hen soup" can be sent directly to the client for display as a filter item; on the contrary, if the icon corresponding to "old hen soup" is obtained, "old hen soup" and the icon corresponding to "old hen soup" can be sent to the client for display together as filter items.

[0125] This embodiment allows users to directly use the filter items as search keywords and obtain more accurate search results, thereby improving the efficiency and accuracy of information search and increasing the richness and aesthetics of information.

[0126] In one embodiment, the prompt templates input into the generative large language model may include a first prompt template and a second prompt template. Based on this, step 103 may be implemented as follows: when the number of search results corresponding to the search content is less than a preset threshold, the server generates search guidance suggestion information using the first prompt template; and when the number of search results corresponding to the search content is greater than or equal to the preset threshold, the server generates search guidance suggestion information using the second prompt template.

[0127] In the above embodiment, the content modules included in the first prompt template and the second prompt template may be the same or different. For example, both prompt templates may include at least one content module of the content modules such as preference dimension, execution steps corresponding to behavior preference dimension, task goal, execution steps of task goal and task output format. At the same time, for the two different search scenarios of zero or few results and broad search terms, the search strategies in the prompt template may have different focuses, that is, the specific contents of the content modules in the two prompt modules may be different. For example, for the search suggestion content in the prompt template, when the number of search results corresponding to the search content is less than the preset number threshold, the corresponding content in the first prompt template may be "Search suggestion: provide search ideas around the search keywords"; when the number of search results corresponding to the search content is greater than or equal to the preset number threshold, the corresponding content in the second prompt template may be "Search suggestion: reason for the suggestion, briefly describe the reason in one sentence". In this way, two different prompt templates with different focuses and similar operation methods can be generated.

[0128] The above embodiment provides two different prompt templates when the number of search results corresponding to the search content is greater than or less than a certain threshold, so that the generated search guidance suggestion information can be more closely integrated with the actual search scenario, thereby improving the accuracy and stability of the search guidance suggestion information.

[0129] In one embodiment, the method of displaying the search guidance suggestion information in step 105 can be implemented in the following manner:

[0130] 1051. The client displays at least one of the search description information, guide words, and filter items in a structured manner, wherein the search description information and / or guide words are displayed through structured text, and the filter items are displayed through structured controls.

[0131] Specifically, after receiving the search guidance suggestion information sent by the server, the client can display at least one of the search description information, guide words and filter items in a structured manner at different locations on the search results page of the client, wherein the search description information and / or guide words can be displayed through structured text, and the filter items can be displayed through structured controls. In this embodiment, the structured manner refers to a method of displaying information according to a preset typesetting method. The structured display method is different from the information display method in a natural state, and can make the displayed information more regular and neat. Moreover, due to the structured typesetting method, the area occupied by each item of information is limited, so it will not take up too much space, nor will it affect the user's normal viewing of search results and other operations.

[0132] The above embodiment displays search guidance suggestion information in a structured manner, allowing it to be neatly displayed on the client. This improves the efficiency and comfort of viewing the search guidance suggestion information, thereby enhancing the search experience. Furthermore, the search guidance suggestion information displayed on the client can effectively address the low hit rate of existing search results, thereby improving the accuracy of information searches.

[0133] In one embodiment, the method of displaying filter items through structured controls in step 1051 can be implemented in the following manner: the client displays at least one filter item through sequentially arranged controls, wherein each control in the sequentially arranged controls is used to display a filter item and to display the search results as search results corresponding to the content in the triggered filter item in response to a triggering operation.

[0134] For example, referring to Figure 5, the client can display multiple filter items, such as "old hen soup", "boiled vegetables" and "scrambled eggs with tomatoes" through multiple controls arranged in sequence. Each filter item can respond to a trigger operation and display the search results as search results corresponding to the content in the triggered filter item. For example, when the user clicks on the control corresponding to "scrambled eggs with tomatoes", the search results page can display the search results corresponding to the keyword "scrambled eggs with tomatoes". It can be understood that the sequential layout referred to in this embodiment can be a horizontal arrangement of multiple controls, a vertical arrangement, etc., without specific limitation.

[0135] This embodiment displays multiple filter items through multiple controls arranged in sequence, which can quickly clarify the user's search intention and does not require the user to re-enter the search content, thereby effectively improving the user's information search efficiency and improving the user's information search experience.

[0136] In one embodiment, the method of displaying the filter items through the structured controls in step 1051 may also be implemented in the following manner: the client displays the text and / or icon corresponding to the filter items through the controls.

[0137] Specifically, referring to Figure 5, the client can display the text, icon, or a combination of text and icon corresponding to the filter items through the control. It is understood that in order to improve the neatness of information display, the information display mode of multiple controls can be the same. For example, multiple controls can only display the text content corresponding to "Old Hen Soup", "Boiled Vegetables", and "Tomato Scrambled Eggs" to ensure the neatness of information display.

[0138] The above embodiment displays the content corresponding to the filter items in the form of text and / or icons, which can make users more aware of the filter items, thereby facilitating users to initiate information search requests by triggering the filter items, thereby improving the efficiency of information search.

[0139] In one embodiment, the filter items are obtained by text matching based on the guide words through preset product names and / or product categories and / or merchant categories.

[0140] Specifically, product names and / or product categories and / or merchant categories can be pre-stored in a filter item vocabulary, or they can be stored in their respective vocabulary according to the original storage method. When generating filter items, by matching the guide words with the names in the filter item vocabulary or with the names in the existing vocabulary one by one, keywords with a high degree of relevance to the guide words can be matched as filter items. In this way, a strong correlation can be established between the guide words and the filter items, thereby improving the consistency of the search guide information and the user's reading experience.

[0141] The above embodiment obtains filtering items by performing text matching between the guide words and the preset product names and / or product categories and / or merchant categories, which can improve the matching degree between the filtering items and the search engine, so that users can accurately search for search results corresponding to the filtering items, thereby improving the user's search accuracy and search efficiency.

[0142] In one embodiment, the filtering items can be product names and / or product categories and / or merchant categories that match the guide words, or product names and / or product categories and / or merchant categories that match the guide words, as well as icons that match the product names and / or product categories and / or merchant categories.

[0143] Specifically, after matching the filtering items through product names and / or product categories and / or merchant categories, the matched keywords can be matched with the names of icons in the preset gallery to obtain icons that match the filtering items. Then, the matched product names and / or product categories and / or merchant categories and the corresponding icons are used as filtering items; in addition, if the icon corresponding to the filtering item is not searched / not found, the matched product names and / or product categories and / or merchant categories can also be directly used as filtering items.

[0144] The above embodiments use product names and / or product categories and / or merchant categories that match the guide words as filter items, and / or use product names and / or product categories and / or merchant categories and their icons that match the guide words as filter items, thereby improving the richness of the filter item information, thereby making it easier for users to view and trigger the filter items to obtain search results corresponding to the filter items, thereby improving search efficiency.

[0145] In one embodiment, the method of displaying the search guidance suggestion information in step 105 may also be implemented in the following manner:

[0146] 1052. The client displays search description information and / or guide words on the first layer of the search results page; and / or the client displays filter items and search results on the second layer of the search results page.

[0147] Among them, the first layer and the second layer of the search result page can be two page layers that overlap with each other on the search result page, for example, the first layer is the background layer, and the second layer is the foreground layer that can be slid and displayed on the first layer, or the second layer is the background layer, and the first layer is the foreground layer that can be slid and displayed on the second layer. In addition, the first layer and the second layer can also be two display areas that do not overlap with each other on the search result page, wherein the placement of the first layer and the second layer can be set according to actual conditions. By placing the search description information and the guide words in one page area for display, the text content can be relatively concentrated, thereby improving the efficiency of users viewing the search guidance suggestion information; by placing the filter items and search results in one page area for display, it can be convenient for users to view the search results corresponding to the information search request. At the same time, it can be convenient for users to refresh the content in the search results by triggering the filter items, thereby improving the efficiency of users viewing the search results and improving the accuracy of information search.

[0148] For example, as shown in Figure 5, the client can display search description information on the background layer of the search results page, such as "Consider using mild foods when you have a cold", "Rich in protein and amino acids, improve immunity", and guide words such as "chicken soup", and then display filter items such as "old hen soup" and search results on the foreground layer of the search results page. By viewing the search description information, the user can clarify the next search scope; by viewing the guide words, the user can clarify the next search intention; by clicking on the filter item, the user can directly view the search results corresponding to the filter item, thereby directly entering the refined search scenario. For example, when the user clicks on the filter item "Tomato Scrambled Eggs", the client can directly display the search content corresponding to "Tomato Scrambled Eggs" on the foreground layer without having to re-initiate the information search request.

[0149] The above embodiment can improve the flexibility of information interaction by displaying multiple categories of information in different areas of the search results page, and can avoid the search guidance suggestion information from blocking the search results, thereby improving the efficiency of users reading the search guidance suggestion information and search results, and improving the efficiency of information search and the interactive experience.

[0150] In one embodiment, the search description information may include search suggestions and guide suggestions, wherein the guide suggestions are associated with the guide words.

[0151] On this basis, after step 1052, the information search method may further include at least one of the following implementation methods:

[0152] 1053. In response to the upward sliding operation of the second layer of the search result page, the client hides part of the guidance suggestions and / or part of the guidance words. When the second layer is displayed on the ceiling, all the guidance suggestions and / or all the guidance words are hidden.

[0153] 1054. In response to the downward sliding operation on the second layer of the search results page, the client displays search description information and / or guide words, and displays at least one historical search content and search guidance suggestion information corresponding to the historical search content.

[0154] Specifically, referring to Figure 5, the client can display search suggestions, guide words, and guide suggestions for guide words on the background layer of the search results page, and display filter items and search results on the foreground layer of the search results page. Through this display method, the user can hide some or all of the guide words and / or guide suggestions when sliding up the foreground page, so as to increase the display area of ​​the foreground page and improve the efficiency of the user in reading the search results. In this scenario, the search suggestions are not blocked, so the user can still view the search suggestions; when the user slides down the foreground layer, the user can view the complete search description information and guide words. In addition, the user can also view the historical search content and the search guide suggestion information corresponding to the historical search content, so as to complete the rapid retracing of historical search records and improve the efficiency of information reading.

[0155] The above embodiment can make the client present different display effects by controlling the up and down sliding of the second layer of the search results page, so that users can flexibly view a variety of information through their own operations, avoiding the search guidance suggestion information from taking up too much space, and improving the efficiency and flexibility of users in viewing the search guidance suggestion information and search results.

[0156] In one embodiment, the search description information may include search suggestions and guidance suggestions, wherein the guidance suggestions are associated with the guide words. Based on this, the method for displaying the search guidance suggestion information in step 105 may also be implemented in the following manner: the client receives the search suggestions and guidance suggestions in real time and dynamically displays the search suggestions and guidance suggestions; after the search suggestions and guidance suggestions are displayed, each guide word and the corresponding guidance suggestion are displayed in segments, with each guide word as the first segment.

[0157] Specifically, the server can send the generated content of the generative large language model to the client word by word in real time as the model outputs content. The client can then dynamically display the search guidance suggestions word by word in a typing manner. This method can speed up the information display process. Moreover, the dynamic display method also helps users notice the generated search guidance suggestions, thereby improving their perception of the search guidance suggestions and thus enhancing the user experience.

[0158] Further, referring to Figure 4, the client can first display search suggestions and guidance suggestions, such as "Consider eating mild foods when you have a cold", "Chicken soup is rich in protein and amino acids, which can improve immunity and anti-inflammatory effects; green vegetables should pay attention to enhancing immunity", etc., and then use the guidance words "chicken soup", "green vegetables" and "tomato" as the beginning of the paragraph to display each guidance word and the guidance suggestions corresponding to the guidance words in segments. In this way, on the one hand, the effect of dynamic display can be enhanced and the user's perception can be improved; on the other hand, the user's comfort in viewing text can be improved. Moreover, due to the length of the text after segmentation, the area where the displayed information is located is limited, which will not affect the user's normal viewing of search results and other operations. In addition, the search suggestions, guidance words and guidance suggestions displayed by the client can effectively make up for the problem of low hit rate of existing retrieval results, thereby improving the accuracy of information search.

[0159] In one embodiment, the information search method may further include the following steps:

[0160] 106. After the search guidance suggestion information is displayed, the client controls the device where the client is located to vibrate.

[0161] Specifically, after the client has fully displayed the search guidance suggestion information, it can control the device where the client is located to vibrate, prompting the user to view relevant content of the AI ​​search. In this way, the user's awareness and participation in the search guidance suggestion information can be improved. In addition, by displaying the search guidance suggestion information, this embodiment can improve search accuracy and address the low hit rate of traditional search results, thereby improving the user's search experience.

[0162] In one embodiment, the information search method may further include at least one of the following methods:

[0163] 501. Before displaying the search guidance suggestion information, the client displays a prompt message indicating that the search guidance suggestion information is being generated, wherein the prompt message is generated according to preset prompt information and / or keywords in the search content.

[0164] Specifically, referring to Figures 2 and 3 , before displaying search guidance suggestions, the client can display a prompt indicating that the search guidance suggestions are being generated, such as "Fewer results detected, casting..." or "Recipes suitable for colds." Because it takes time for the generative large language model to generate and send search guidance suggestions to the client, displaying this prompt allows users to understand the relevant functions of AI search in advance, thereby increasing their awareness of the search guidance suggestions.

[0165] 502. Before displaying the search guidance suggestion information, the client dynamically displays the information loading icon.

[0166] Specifically, referring to Figures 2 and 3, the client can dynamically display an information loading icon before displaying the search guidance suggestion information. The information loading icon can be displayed together with the prompt information or displayed alone. By dynamically displaying the information loading icon, the perception of the artificial intelligence search function can be improved.

[0167] 503. The client displays the search content and artificial intelligence icon in the information search box.

[0168] Specifically, referring to Figures 2 to 5 , the client can display the search content and an AI icon in the information search box. The AI ​​icon is used to identify the information search request corresponding to the search content as a preset AI search request. By displaying the AI ​​icon, the user is prompted that the information search request initiated is a preset AI search request, thereby enhancing the user's awareness of the AI ​​search function.

[0169] 504. In response to detecting a network interruption, the client displays an information loading control, wherein the information loading control is used to re-receive and display the search guidance suggestion information in response to a triggering operation.

[0170] Specifically, after sending an information search request, the client may be interrupted in receiving search guidance suggestion information due to network issues. The client can monitor the information reception status or network status in real time and, if a network or information reception interruption is detected, display an information loading control. After the user clicks the information loading control, the client can re-receive the search guidance suggestion information sent by the server, either from the beginning or intermittently, which is not specifically limited in this embodiment.

[0171] The above embodiments can enhance users' awareness and participation in artificial intelligence search functions through a variety of information display methods, thereby enhancing users' search experience.

[0172] In one embodiment, as a refinement and expansion of the specific implementation of the above embodiment, in order to fully illustrate the implementation process of this embodiment, an information search method is provided, as shown in Figures 6 and 7. The method can be implemented in the following manner: when the user enters the search content (the pre-search content in this case) in the information search box of the client, the client can respond to the information input operation, display the search association page, and send the pre-search content to the server. The server can identify the keywords in the pre-search content and determine whether the keywords in the pre-search content match the keywords in the artificial intelligence offline vocabulary. If the keywords in the vocabulary are hit, the artificial intelligence entry can be encapsulated on the server side through artificial intelligence terms and artificial intelligence mode tags, and the artificial intelligence entry can be sent to the client. The client can display the artificial intelligence entry in the search association page, and at the same time, it can also display other related terms related to the keywords in the search content. At this time, the client can execute different search methods according to the different operations performed by the user.

[0173] Furthermore, if the user directly sends the search content and the search content does not meet the preset conditions, or selects a common term to send, the client can send the information search request to the server's search engine to obtain the search results corresponding to the information search request and display the search results on the search results page; if the user sends the search content and the search content meets the preset conditions, or clicks on the artificial intelligence entrance, the client can, on the one hand, send the information search request to the server's search engine to obtain the search results corresponding to the information search request and then display the search results on the search results page; on the other hand, the client can send the artificial intelligence term and the information search request to the server's generative large language model. The server can then start the generative large language model, which can select a corresponding prompt template based on the number of search results corresponding to the search content and perform information prediction such as expanded keywords to generate search guidance suggestion information. The generated search guidance suggestion information is then packaged and sent to the client for rendering and display. At this time, the user can view the search guidance suggestion information corresponding to the artificial intelligence term through the client, and at the same time, can also view the search results corresponding to the search content.

[0174] By automatically triggering preset AI search requests in response to user search operations, this method can effectively improve user search efficiency and the accuracy of search results. Furthermore, this method does not affect the display of normal search results or cause excessive disruption to users, and can make search results more relevant to users' actual search needs, thereby effectively enhancing the user's search experience.

[0175] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, storage, and display, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. In addition, the numbers corresponding to the various steps in the above embodiments serve only as identifiers and do not limit the order in which the steps are executed. The order in which the steps are executed in each embodiment can be set according to actual circumstances.

[0176] Furthermore, as a specific implementation of the method shown in FIG. 1 to FIG. 7 , this embodiment provides an information search device, as shown in FIG. 8 , which includes: a search request response module 61 and a guidance suggestion display module 62 , wherein:

[0177] The search request response module 61 is configured to send an information search request in response to receiving the information search request;

[0178] The guidance suggestion display module 62 can be used to receive and display search guidance suggestion information and search results, wherein the search guidance suggestion information is generated based on the search content through a generative large language model, and the search results are obtained through the information search request; the search guidance suggestion information includes at least one of search description information, guidance words and filter items; the guidance words and the filter items are associated with each other, and the filter items are used to display the search results as search results corresponding to the content in the filter items.

[0179] Furthermore, as a specific implementation of the method shown in Figures 1 to 7, this embodiment provides an information search device, as shown in Figure 9, which includes: a search request response module 71, a guidance suggestion generation module 72 and a guidance suggestion sending module 73.

[0180] The search request response module 71 is configured to, in response to receiving an information search request, obtain search content corresponding to the information request;

[0181] A guidance suggestion generating module 72 may be configured to generate search guidance suggestion information based on the search content using a generative large language model, and obtain search results corresponding to the information search request, wherein the search guidance suggestion information includes at least one of search description information, a guidance word, and a filter item, wherein the guidance word and the filter item are associated with each other;

[0182] The guidance suggestion sending module 73 is used to send the search guidance suggestion information and the search results to the client, and the filter item is used to display the search results as search results corresponding to the content in the filter item.

[0183] It should be noted that for other corresponding descriptions of the functional units involved in the information search device provided in this embodiment, reference can be made to the corresponding descriptions in Figures 1 to 7, which will not be repeated here.

[0184] Based on the above methods shown in FIG. 1 to FIG. 7 , this embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the above information search methods shown in FIG. 1 to FIG. 7 are implemented.

[0185] Based on this understanding, the technical solution of the present application can be embodied in the form of a software product. The software product to be identified can be stored in a non-volatile storage medium (which can be a CD-ROM, USB flash drive, mobile hard disk, etc.), including a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods described in each implementation scenario of the present application.

[0186] Based on the above-mentioned methods as shown in Figures 1 to 7, and the information search device embodiments shown in Figures 8 and 9, in order to achieve the above-mentioned purpose, as shown in Figure 10, this embodiment also provides a computer device for information search, which can be specifically a personal computer, server, smart phone, tablet computer, smart watch, or other network device, etc. The computer device includes a storage medium and a processor; the storage medium is used to store computer programs and operating systems; the processor is used to execute computer programs to implement the above-mentioned methods as shown in Figures 1 to 7.

[0187] Optionally, the computer device may further include an internal memory, a communication interface, a network interface, a camera, a radio frequency (RF) circuit, a sensor, an audio circuit, a Wi-Fi module, a display, an input device such as a keyboard, etc. Optionally, the communication interface may further include a USB interface, a card reader interface, etc. The network interface may optionally include a standard wired interface, a wireless interface (such as a Wi-Fi interface), etc.

[0188] Those skilled in the art will understand that the information search computer device structure provided in this embodiment does not constitute a limitation on the computer device, and may include more or fewer components, or a combination of certain components, or different component arrangements.

[0189] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages the computer device hardware and the software resources to be identified, supporting the execution of the information processing program and other software and / or programs to be identified. The network communication module is used to enable communication between components within the storage medium and with other hardware and software in the information processing computer device.

[0190] Through the description of the above implementation methods, those skilled in the art can clearly understand that the present application can be implemented with the help of software plus the necessary general hardware platform, or it can be implemented through hardware. By applying the technical solution of the present application, first responding to the information search request initiated by the user, the information search request is sent to the server so that the server obtains the search content corresponding to the information search request, and then, based on the search content, through the generative large language model, a search guidance suggestion information covering a variety of information such as search description information, guide words and filter items is generated, and the search results corresponding to the information search request are obtained. Finally, the generated search guidance suggestion information and the obtained search results are sent to the client for display. Compared with the prior art, the above method can provide users with more accurate and personalized search results, and present the search results to users in a more explainable way, thereby improving the accuracy of the search results and enhancing the user's information search experience.

[0191] Those skilled in the art will understand that the accompanying drawings are only schematic diagrams of a preferred implementation scenario, and the modules or processes in the accompanying drawings are not necessarily required to implement the present application. Those skilled in the art will understand that the modules in the devices in the implementation scenario can be distributed in the devices of the implementation scenario according to the implementation scenario description, or can be changed accordingly and located in one or more devices different from the implementation scenario. The modules of the above-mentioned implementation scenario can be combined into one module, or can be further split into multiple sub-modules.

[0192] The serial numbers of the above application are for descriptive purposes only and do not represent the advantages or disadvantages of the implementation scenarios. The above disclosure only discloses several specific implementation scenarios of the present application, but the present application is not limited thereto. Any changes that can be conceived by those skilled in the art should fall within the scope of protection of the present application.

Claims

1. An information search method, characterized in that, The method includes: In response to receiving an information search request, the client sends the information search request to the server; In response to receiving the information search request, the server obtains the search content corresponding to the information search request; Based on the search content, the server generates search guidance suggestion information through a generative large language model and obtains search results corresponding to the information search request, where the search guidance suggestion information includes at least one of search description information, guiding words, and filtering options, and the guiding words and the filtering options are interrelated; The server sends the search guidance suggestion information and the search results to the client; The client receives and displays the search guidance suggestion information and the search results, and the filtering option is used to display the search results as search results corresponding to the content in the filtering option.

2. The method according to claim 1, wherein The client receiving and displaying the search guidance suggestion information includes: The client displays at least one of the search description information, the guiding words, and the filtering options in a structured manner, where the search description information and / or the guiding words are displayed in structured text, and the filtering options are displayed in structured controls.

3. The method according to claim 2, wherein The client displaying the filtering options through the structured controls includes: The client displays at least one of the filtering options through sequentially arranged controls, where each control in the sequentially arranged controls is used to display one of the filtering options and is used to, in response to a trigger operation, display the search results as search results corresponding to the content in the triggered filtering option.

4. The method according to claim 2 or 3, characterized in that, The client displaying the filtering options through the structured controls further includes: The client displays the text and / or icon corresponding to the filtering option through the control.

5. An information search method, characterized in that, The method includes: In response to receiving an information search request, sending the information search request; Receiving and displaying search guidance suggestion information and search results, where the search guidance suggestion information is generated through a generative large language model based on the search content, and the search results are obtained through the information search request; the search guidance suggestion information includes at least one of search description information, guiding words, and filtering options; the guiding words and the filtering options are interrelated, and the filtering option is used to display the search results as search results corresponding to the content in the filtering option.

6. An information search method, characterized in that, The method includes: In response to receiving an information search request, obtaining the search content corresponding to the information request; Based on the search content, generating search guidance suggestion information through a generative large language model and obtaining search results corresponding to the information search request, where the search guidance suggestion information includes at least one of search description information, guiding words, and filtering options, and the guiding words and the filtering options are interrelated; Sending the search guidance suggestion information and the search results to the client, and the filtering option is used to display the search results as search results corresponding to the content in the filtering option.

7. An information search device, characterized in that, The device includes: A search request response module, configured to send the information search request in response to receiving an information search request; A guidance suggestion display module, configured to receive and display search guidance suggestion information and search results, wherein the search guidance suggestion information is generated by a generative large language model based on the search content, and the search results are obtained through the information search request; the search guidance suggestion information includes at least one of search description information, guiding words, and filtering options; the guiding words and the filtering options are correlated with each other, and the filtering options are used to display the search results as search results corresponding to the content in the filtering options. The device includes:

8. An information search device, characterized in that, A search request response module, configured to obtain the search content corresponding to the information request in response to receiving an information search request; A guidance suggestion generation module, configured to generate search guidance suggestion information through a generative large language model based on the search content, and obtain search results corresponding to the information search request, wherein the search guidance suggestion information includes at least one of search description information, guiding words, and filtering options, and the guiding words and the filtering options are correlated with each other; A guidance suggestion sending module, configured to send the search guidance suggestion information and the search results to the client, and the filtering options are used to display the search results as search results corresponding to the content in the filtering options. The computer program, when executed by a processor, implements the steps of the method according to any one of claims 1 to 6.

9. A storage medium having a computer program stored thereon, characterized in that, The computer program, when executed by a processor, implements the steps of the method according to any one of claims 1 to 6.

10. A computer device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, ​

Images