Text search method and device

Abstract

The embodiment of the invention provides a text search method and device, and the method comprises the steps: responding to a text selection operation, and determining target text information from a current display page of a target document; identifying and determining a target vocabulary and a corresponding target intention type based on the target text information; and performing information search based on at least one of a plurality of set search links, the target vocabulary and the corresponding target intention type to obtain target associated information corresponding to the target vocabulary. According to the scheme, the actual intention demand of the user can be recognized in combination with the text selected by the user, the text search matched with the actual intention demand is performed through the set search link, the accurate search result is provided, the actual query demand of the user is met, and the user experience is improved.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a text search method and apparatus. Background Technology

[0002] With the continuous development of information technology, digital office work has become increasingly prevalent in people's learning, work, and life, meeting diverse user needs. For example, users can read, edit, and extract information from electronic documents through different terminal devices to satisfy various office requirements. In order to better understand text content or other office needs, users need to perform related searches on words to obtain relevant definitions, examples, and other related information. With the advancement of digital office technology, related technologies have provided convenient and intelligent tools for word lookup, facilitating users' text search needs.

[0003] Text search in related technologies specifically involves searching the internet for specific terms, using electronic resources such as electronic dictionaries and encyclopedias to find basic definitions. However, due to the wide variety and inconsistent quality of information sources on the internet, accurate search results cannot be provided, deviating from the user's actual query needs and affecting the user experience. Summary of the Invention

[0004] This application provides a text search method and apparatus to solve the problem in related technologies that cannot provide accurate search results, deviate from the user's actual query needs, and affect the user experience. It can identify the user's actual intent by combining the text selected by the user, and perform text search that matches the actual intent through a set search link, providing accurate search results that fit the user's actual query needs and improve the user experience.

[0005] In a first aspect, embodiments of this application provide a text search method, the method comprising: In response to a text selection operation, determine the target text information from the currently displayed page of the target document; Based on the target text information, target words and corresponding target intent types are identified and determined; Information is searched based on at least one of the multiple search links set, the target vocabulary and the corresponding target intent type to obtain the target association information corresponding to the target vocabulary.

[0006] Secondly, embodiments of this application also provide a text search device, including: The text information determination module is configured to determine the target text information from the currently displayed page of the target document in response to a text selection operation. The vocabulary intent determination module is configured to identify and determine target words and corresponding target intent types based on the target text information. The association information determination module is configured to perform information search based on at least one of a plurality of set search links, the target word and the corresponding target intent type, to obtain the target association information corresponding to the target word.

[0007] Thirdly, embodiments of this application also provide a text search device, the device comprising: One or more processors; Storage device, configured to store one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the text search method described in the embodiments of this application.

[0008] Fourthly, embodiments of this application also provide a non-volatile storage medium for storing computer-executable instructions, which, when executed by a computer processor, are configured to perform the text search method described in embodiments of this application.

[0009] In this embodiment, by responding to the user's text selection operation, the text information of interest to the user can be accurately extracted, eliminating interference from irrelevant content in the document. Based on the recognition of the target text information, the most crucial target words can be extracted, and the target intent type can be determined, making subsequent search results more aligned with the user's actual needs. By using a set search path based on the target words and target intent type, the associated information of the target words can be accurately obtained, satisfying diverse search needs. The above solution can combine the user's selected text to identify the user's actual intent needs, and perform text searches matching these actual intent needs through a set search path, providing accurate search results that fit the user's actual query requirements and improve the user experience. Attached Figure Description

[0010] Figure 1 A flowchart illustrating a text search method provided in this application embodiment; Figure 2 A schematic diagram illustrating a process of generating target-related information using an applied text search method, as provided in an embodiment of this application; Figure 3 A flowchart illustrating a text search method that includes a process for determining target association information, provided as an embodiment of this application; Figure 4 A flowchart for generating a first search result based on a first search link is provided as an embodiment of this application; Figure 5A flowchart for generating a second search result based on a second search link is provided as an embodiment of this application; Figure 6 A flowchart for determining first partial text information is provided in an embodiment of this application; Figure 7 A flowchart for generating a third search result based on a third search link is provided as an embodiment of this application; Figure 8 A flowchart illustrating a specific method for generating a third search result based on directory matching results, provided in this application embodiment; Figure 9 A flowchart for generating a fourth search result based on a fourth search link is provided in this application embodiment; Figure 10 This is a schematic diagram illustrating the selection of multiple search links and the integration of their respective search results, as provided in an embodiment of this application. Figure 11 A flowchart illustrating the integration of search results corresponding to various target search links is provided in this application embodiment. Figure 12 A structural block diagram of a text search device provided in an embodiment of this application; Figure 13 This is a schematic diagram of the structure of a text search device provided in an embodiment of this application. Detailed Implementation

[0011] The embodiments of this application will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely illustrative of the embodiments of this application and are not intended to limit the scope of the embodiments. Furthermore, it should be noted that, for ease of description, only the parts relevant to the embodiments of this application are shown in the accompanying drawings, not the entire structure.

[0012] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0013] The text search method provided in this application is used to identify the target words and target intent types that a user needs to query, and to perform information search through a set search link to provide target-related information corresponding to the target words. Specifically, it can be applied to scenarios where users read, edit, and extract information from electronic documents, but this application does not limit its application to these scenarios. This application aims to provide a text search method and apparatus to solve the problem in related technologies that cannot provide accurate search results, deviate from the user's actual query needs, and affect the user experience.

[0014] The text search method provided in this application embodiment can be executed by a computer device. The computer device refers to any electronic device with data computing, processing and storage capabilities, such as mobile phones, PCs (Personal Computers), tablet computers and other terminal devices, or servers and other devices. This application embodiment does not limit the scope of the computer device.

[0015] Figure 1 A flowchart of a text search method provided in an embodiment of this application is shown below. Figure 1 As shown, this text search method includes the following steps: Step S101: In response to the text selection operation, determine the target text information from the currently displayed page of the target document.

[0016] The target document can be an electronic document that the user is currently interacting with, specifically a locally stored document, an online document, or an application-embedded document, etc., which are not limited in this application. The currently displayed page can be the page content of the target document currently displayed on the screen of the user's terminal device. For example, if the user is currently browsing a 50-page electronic document and page 15 is currently displayed on the screen, then the currently displayed page is the visible content of page 15. The text selection operation can be an operation in which the user selects a specific range of text content on the currently displayed page of the target document through interactive methods such as mouse dragging, long-press selection on the touch screen, or selection using a shortcut key. Since different document types of target documents have different corresponding text information extraction methods, the specific implementation process of determining the target text information from the currently displayed page of the target document in response to the text selection operation is as follows: The current page is parsed to determine the page text information based on the document type of the target document; the target text information is determined from the page text information based on the area coordinate information corresponding to the text selection operation.

[0017] The document type can be a streaming document, a layout document, etc. A streaming document may have content that is not bound to a fixed page layout, but rather dynamically adjusts its layout based on the container size (e.g., document window size, device screen width, etc.). A layout document may have content bound to a fixed page layout, maintaining consistency with the original design layout across different devices or window sizes. It should be noted that parsing the currently displayed page specifically involves converting its raw page data into structured page text information, which may include text content, location-related content, etc. For example, in a streaming document, the parsed page text information may be text character encoding and character position mapping relationships. In a layout document, the parsed page text information may be text block coordinates and character encodings within the block. Region coordinate information may be the pixel position coordinates corresponding to the selected region in a text selection operation. In one embodiment, if the target document is a streaming document, based on the pixel range selected by the region coordinate information corresponding to the text selection operation, the text character encodings within that pixel range in the page text information can be determined. Combining the order in which the text character encodings appear and the character position mapping relationships, the text characters are arranged sequentially to obtain the target text information. In one embodiment, if the target document is a layout document, based on the pixel range selected by the region coordinate information corresponding to the text selection operation, multiple text blocks located within that pixel range in the page text information can be determined. According to the projection order of the text blocks in each row on the X-axis, or the projection order of the text blocks in each column on the Y-axis, the multiple text blocks can be arranged according to the actual reading order. Extracting the text characters from the arranged multiple text blocks yields the target text information. Therefore, text information extraction can be adapted to different document types, accurately locating the initial reference content provided by the user.

[0018] Step S102: Identify and determine the target words and corresponding target intent types based on the target text information.

[0019] The target text information can be the specific text content selected by the text selection operation on the currently displayed page. Identifying this target text information can specifically involve using an artificial intelligence model or preset matching rules to perform semantic content analysis of the target text information, and extracting or matching target words and corresponding target intent types. In one embodiment, the specific implementation process for identifying and determining target words and corresponding target intent types based on target text information is as follows: The sixth prompt word information is generated based on the target text information and the fourth prompt word template. The sixth prompt word information is then input into the artificial intelligence model to obtain at least one candidate word and the candidate intent type corresponding to each candidate word. In response to the information selection operation, a target word and its corresponding target intent type are determined from at least one candidate word and the candidate intent type corresponding to each candidate word.

[0020] The fourth prompt word template can be a predefined structured text framework used to convert target text information into a prompt word format suitable for input into an artificial intelligence model. Specifically, it can include fixed prompts and variable placeholders, the latter of which can be used to populate the target text information. For example, the fourth prompt word template is as follows: Please analyze the following input text to determine the target words the user is most likely looking for, and the type of intent the user intends to use to search for those target words (e.g., definition, case study, application, background, principle, steps, comparison, etc.). The output format must be: <anchor> Target vocabulary< / anchor> <intent> Intent type< / intent> Requirements: 1. Choose the most crucial and core vocabulary possible. 2. The intent type should be summarized with a concise category term, such as: definition, case study, application, background, principle, steps, comparison, etc. 3. Do not output additional explanations, do not add serial numbers, and do not attach other content. 4. If the intent is unclear, output "Unclear". 5. If multiple target words or multiple intent types are identified, output them in order of probability. Input text: "+anchorStr

[0021] Here, "+anchorStr" is a placeholder corresponding to the target text information. The sixth prompt word information is generated based on the target text information and the fourth prompt word template. Specifically, the target text information can be filled into the placeholder position corresponding to the fourth prompt word template to obtain the sixth prompt word information. This sixth prompt word information can be input into an artificial intelligence model to predict the target vocabulary and the corresponding target intent type. The artificial intelligence model can be a large language model such as GPT4, DeepseekV3, or Qwen2.5, and is not limited to this in this application. In one embodiment, after the sixth prompt word information is input into the artificial intelligence model, if the model outputs "unclear" and does not show a target vocabulary or target intent type, then the corresponding message "Intent comprehension failed; please clarify the content to be queried" can be displayed on the current screen to prompt the user to reconfirm the content to be queried. In one embodiment, after inputting the sixth prompt word information into the artificial intelligence model, at least one candidate vocabulary and a candidate intent type corresponding to each candidate vocabulary can be obtained. If both the number of candidate vocabulary and the number of candidate intent types are single, then the candidate vocabulary can be directly identified as the target vocabulary, and the candidate intent type can be identified as the target intent type. If there are multiple candidate words and multiple candidate intent types, then options corresponding to these multiple candidate words and multiple candidate intent types can be provided in the current display interface to receive the user's information selection operation, and to determine the target word and the corresponding target intent type from the multiple candidate words and multiple candidate intent types. Optionally, this information selection operation can be achieved through clicking, checking, voice commands, etc., which are not limited in this application. Thus, the target text information can be analyzed by an artificial intelligence model to accurately extract words and intent types that meet the user's requirements, thereby improving processing efficiency.

[0022] In one embodiment, the specific implementation process for identifying target words and corresponding target intent types based on target text information is as follows: The target text information is segmented into multiple lexical units. First, meaningless function words, adjectives, and other irrelevant lexical units are removed based on preset grammatical rules, retaining only meaningful proper nouns, terms, and other candidate words. Then, using preset intent matching rules, the sentence structure or auxiliary words in the sentence containing the candidate words are matched to determine the corresponding candidate intent type. Finally, if there are multiple candidate words and multiple candidate intent types, options corresponding to these multiple candidate words and multiple candidate intent types can be provided on the current display interface to allow the user to select and confirm the final target word and corresponding target intent type.

[0023] Step S103: Based on at least one of the multiple search links set, the target word and the corresponding target intent type, perform information search to obtain the target association information corresponding to the target word.

[0024] The search link can be a pre-configured standardized processing flow for target-related information corresponding to the target keywords. The specific configuration can be tailored to the feasible search paths and business requirements of the actual application scenario. Target-related information can include relevant definitions, examples, and background information for the target keywords, determined by the aforementioned target intent type. Information is searched based on at least one of the configured search links, the target keywords, and the corresponding target intent type. Specifically, for the target intent type, the search link's built-in standardized processing flow can be used to analyze or retrieve the target keywords from different search sources to obtain the target-related information corresponding to the target keywords.

[0025] For example, starting from the search source, search links based on network resources and local resources can be set up accordingly. The data for the network resource-based search link comes from external web page content, while the data for the local resource-based search link comes from locally stored document content, such as the target document or a set document database. In one embodiment, the network resource-based search link can extract associated content from multiple recorded open-source web pages based on the target keywords and corresponding target intent types, and then clean and integrate the associated content to obtain the target association information corresponding to the target keywords. In one embodiment, the local resource-based search link can parse and retrieve the target association information corresponding to the target keywords from the currently opened target document. Furthermore, the local resource-based search link can be divided into multiple different links based on different content search scopes. The content search scope can be the entire document content of the target document or a portion of the document content. The partial document content can be the document content corresponding to the directory item after matching the target keywords with the target document's table of contents, or it can be the document content determined with reference to the user-input page number range. In one embodiment, each set search link can be executed in parallel, and the search results corresponding to each search link can be integrated to obtain target association information. In another embodiment, a selection control for multiple set search links can be provided on the current display interface to receive the user's link selection operation and determine which one or more of the multiple search links to use. Optionally, the search link configuration last used by the user can be selected by default to quickly provide target association information.

[0026] Optionally, the number of target words is at least two. It should be noted that the number of target text information can be one or more. These multiple target words can be located in the same target text information or in different target text information; this application does not impose any limitations on this. In addition, the target intent types corresponding to different target words can be the same, partially different, or completely different; this application does not impose any limitations on this either.

[0027] In one embodiment, the number of target words is at least two, and at least two target words correspond to the same target intent type. The specific implementation process for obtaining target association information corresponding to the target words by performing information search based on at least one of multiple set search links, the target words, and the corresponding target intent types is as follows: Information is searched based on at least one or at least two of the target words and their corresponding target intent types from multiple search links to obtain integrated target association information corresponding to at least two of the target words.

[0028] In one embodiment, the number of target words is at least two, and the target intent types corresponding to at least two target words are partially or completely different; the specific implementation process of performing information search based on at least one of the multiple set search links, the target words, and the corresponding target intent types to obtain the target association information corresponding to the target words is as follows: Information is searched based on at least one or at least two of the target words and their corresponding target intent types from multiple search links to obtain integrated target association information corresponding to at least two of the target words.

[0029] Specifically, information is searched based on at least one and at least two of the target words from multiple search links and their corresponding target intent types to obtain integrated target association information corresponding to at least two of the target words, including: In response to the link selection operation, at least one target search link is determined from the multiple search links set.

[0030] The link selection operation allows users to select one or more target search links from multiple set search links through interface interaction (e.g., clicking a button, checking an option, or selecting from a drop-down menu). Each target search link can execute its corresponding processing flow to match multiple determined target words with the corresponding target intent type of information search.

[0031] For each target word, the following processing is performed: according to the processing flow corresponding to each target search link, the target word is searched for information matching the target intent type to obtain the search results corresponding to each target search link; the search results corresponding to each target search link are integrated to obtain the target association information corresponding to the target word.

[0032] The target association information corresponding to each target word is integrated to obtain the integrated target association information.

[0033] Since there are multiple target words, each target search link can adopt a variety of processing strategies for these multiple target words.

[0034] In one specific implementation, the target search link can perform information searches for each target word in sequence or in parallel to match its corresponding target intent type. Thus, each target word can independently obtain its corresponding search results. The search results corresponding to each target word can be summarized to obtain the search results corresponding to the target search link.

[0035] In another specific implementation, for example, multiple target terms include a first target term and a second target term. The target search link can first perform an information search on the first target term to match its corresponding target intent type, obtaining a first association result. This first association result can serve as supplementary information for information searching on the next target term. Next, the target search link can combine the second target term and the first association result to obtain combined information, and then perform an information search on this combined information to match the target intent type corresponding to the second target term, obtaining a second association result. This second association result can be considered the search result of the target search link.

[0036] It should be noted that if a user selects multiple target search links, for the same target keyword, the search results corresponding to the target keyword from different target search links can be deduplicated, sorted, or merged to obtain the target association information corresponding to the target keyword.

[0037] After obtaining the target association information corresponding to each target word, there are multiple integration methods to choose from to process and integrate the target association information.

[0038] In one specific implementation, the integration method can be to filter duplicates from the target association information corresponding to each target word to obtain integrated target association information. Specifically, the target association information corresponding to each target word can be hashed to obtain corresponding hash values, the hash values ​​can be compared, and target association information with the same hash value can be filtered out. Alternatively, the target association information corresponding to each target word can be compared string by string to identify and filter duplicate target association information. Finally, the filtered target association information can be listed and displayed as integrated target association information.

[0039] In one specific implementation, the integration method can be to semantically fuse the target association information corresponding to multiple target words to obtain integrated target association information. Specifically, the multiple target association information can be filled into a preset prompt word template to obtain prompt word information. The preset prompt word template contains relevant guiding words with pre-set semantic fusion rules. Inputting the prompt word information into a set artificial intelligence model can obtain the model fusion result, which can be regarded as the integrated target association information.

[0040] In one specific implementation, the integration method can be to obtain integrated target relationship information by ranking the relevance of the target association information corresponding to multiple target words. Specifically, the target association information corresponding to each target word can be segmented, stop words removed, and stemmed to obtain keyword information. A pre-trained model (such as BERT, Word2Vec, etc.) is used to convert the keyword information into a high-dimensional semantic vector. A target association information is selected, and the semantic similarity between the other target association information and the selected target association information is calculated. The target association information is then ranked from high to low according to the semantic similarity to obtain the integrated target association information. Figure 2 This application provides a schematic diagram illustrating a process for generating target-related information using an applied text search method, as shown in the embodiments of this application. Figure 2 As shown, in response to a text selection operation, target text information 202 is determined in the currently displayed page 201 of the target document. Then, by identifying the target text information 202, target vocabulary 203 (exemplarily given as "natural resource data" in the figure) and the corresponding target intent type 204 (exemplarily given as "case" in the figure) can be determined. Finally, based on at least one of the set multiple search links 205, the target vocabulary 203, and the corresponding target intent type 204, information search is performed to obtain target association information 206 corresponding to the target vocabulary 203, wherein the target association information 206 is in Figure 2 It is displayed in a card format.

[0041] As described above, by responding to the user's text selection, the system can accurately extract text information of interest to the user while eliminating interference from irrelevant content in the document. Based on the recognition of the target text information, the system can extract the most crucial target words and determine the target intent type, making subsequent search results more aligned with the user's actual needs. By using a set search path based on target words and target intent type, the system accurately obtains related information about the target words, satisfying diverse search needs. This solution combines the user's selected text with the recognition of the user's actual intent, and uses a set search path to perform text searches matching that intent, providing accurate search results that fit the user's actual query needs and improve the user experience.

[0042] Figure 3 A flowchart of a text search method including a process for determining target association information is provided for embodiments of this application, as shown below. Figure 3 As shown, this text search method includes the following steps: Step S301: In response to the text selection operation, determine the target text information from the currently displayed page of the target document.

[0043] Step S302: Identify and determine the target words and corresponding target intent types based on the target text information.

[0044] Step S303: In response to the link selection operation, determine at least one target search link from the set multiple search links.

[0045] The link selection operation allows users to choose one or more target search links from multiple set search links through interface interaction (e.g., clicking a button, checking an option, or selecting from a drop-down menu), so that users can choose information sources or search strategies.

[0046] Step S304: According to the processing flow corresponding to each target search link, perform information search on the target words to match the target intent type, and obtain the search results corresponding to each target search link.

[0047] Each target search link incorporates a standardized processing flow to ensure consistency and repeatability in the search process. Since different target intent types (e.g., definition, case study, purpose, background, principle, steps, comparison, etc.) have different requirements for related information on target terms, each target search link can match the target intent type for information searching, ensuring a high degree of match between search behavior and user needs. Each target search link can then provide the corresponding search results after executing the search.

[0048] In one embodiment, the aforementioned identified at least one target search link includes a first search link. Figure 4A flowchart for generating a first search result based on a first search link is provided as an embodiment of this application, such as... Figure 4 As shown, the target keywords are matched with target intent types for information retrieval according to the processing flow corresponding to each target search link, and the search results for each target search link are obtained. The specific steps include: Step S401: When the target search link is the first search link, perform network retrieval based on the target keywords to obtain relevant content information.

[0049] The first search link can be a search link based on network resources. Specifically, this network retrieval can use regular expressions or keyword matching to filter text content containing the target keywords from multiple pre-configured network pages, and then perform deduplication, cleaning, and formatting processing on all the filtered text content to obtain relevant content information. Of course, other network search methods are also possible, and this application does not limit them. Optionally, the search results for each network page can be scored based on the information quality of the network pages and the frequency or position of the target keywords on each network page, and the search results of the preset number of network pages with the highest scores can be retained as relevant content information.

[0050] Step S402: Generate first prompt information based on target vocabulary, target intent type, relevant content information and first prompt word template.

[0051] The first prompt word template can be used to convert target vocabulary, target intent type, and related content information into a prompt word format suitable for input into an artificial intelligence model. For example, the first prompt word template is as follows: You will receive the following three types of information: target keywords, target intent type, and relevant content information obtained from the web search. Based on the relevant content information from the web search, combined with the target keywords and target intent type, please generate the first search result and output it in the following format: <result> First search result< / result> The rules are as follows: The first search result should strictly meet the user's intent. For example, if the intent is "definition," output an accurate and concise definition. If the intent is "case study," output relevant case studies. If the intent is "purpose," explain the purpose. If the search results come from multiple sources, they need to be integrated, removing irrelevant or duplicate content to ensure the results are coherent, concise, and accurate. Do not add explanations, descriptions, serial numbers, quotation marks, or extra text to the output. If the first search result is irrelevant to the intent, <result>The tag outputs "Unclear". The following are the target keywords: " + anchor + " The following are the target intent types: " + intent + " The following are relevant content information from web searches: " + IntelContext.

[0052] Wherein, "+ anchor +" is a placeholder corresponding to the target vocabulary, "+ intent +" is a placeholder corresponding to the target intent type, and "+ IntelContext" is a placeholder corresponding to the relevant content information. First prompt word information is generated based on the target vocabulary, target intent type, relevant content information, and first prompt word template. Specifically, the target vocabulary, target intent type, and relevant content information can be filled into the placeholder positions corresponding to the first prompt word template to obtain the first prompt word information. This first prompt word information can be input into an artificial intelligence model to output the final first search result. The artificial intelligence model can be a large language model such as GPT4, DeepseekV3, or Qwen2.5, and is not limited in this application. In one embodiment, after the first prompt word information is input into the artificial intelligence model, if the output result of the artificial intelligence model is "unclear," then the current display interface can display "Network search failed, please try other search links" to prompt the user to reselect another search link.

[0053] Step S403: Input the first prompt word information into the artificial intelligence model to obtain the first search result corresponding to the first search link.

[0054] Therefore, the first search link can extract relevant content information that matches the target keywords from web page resources, and use the processing power of artificial intelligence models to further filter and integrate the relevant content information to obtain the first search result that fits the target intent type.

[0055] In one embodiment, the aforementioned identified at least one target search link includes a second search link. Figure 5 A flowchart for generating a second search result based on a second search link is provided as an embodiment of this application, such as... Figure 5 As shown, the target keywords are matched with target intent types for information retrieval according to the processing flow corresponding to each target search link, and the search results for each target search link are obtained. The specific steps include: Step S404: If the target search link is the second search link, parse the target document to determine the full text information.

[0056] The second search link can be a search link based on local resources, and its content search scope can be the entire document content of the target document. By parsing the target document, the original document data can be transformed into structured full text information. The specific parsing process can be referred to the relevant description of determining the target text information in the foregoing embodiments, and will not be repeated here.

[0057] Step S405: Determine the first local text information that meets the second similarity condition from the full text information based on the target vocabulary, and generate the second prompt word information based on the target vocabulary, target intent type, first local text information and second prompt word template.

[0058] The second similarity condition can be used as a quantitative rule to filter local text information that matches the target word from the full text information. Specifically, the first local text information that satisfies the second similarity condition is determined from the full text information based on the target word. This can be achieved by performing content similarity judgment between the target word and the full text information according to the similarity constraints of the second similarity condition, and extracting the text content that satisfies the second similarity condition to obtain the first local text information. In one embodiment, Figure 6 A flowchart for determining first partial text information is provided as an embodiment of this application, such as... Figure 6 As shown, the method for determining the first local text information that satisfies the second similarity condition from the full text information based on the target vocabulary includes the following steps: Step S4051: Segment the full text information to obtain multiple text segments, vectorize each text segment to obtain a first text vector, and vectorize the target words to obtain a second text vector.

[0059] The text fragments can be local text units obtained by segmenting the full text information according to preset rules. In one embodiment, the target document can be segmented into multiple text fragments according to its original paragraph structure. In another embodiment, the full text information can be segmented into multiple text fragments according to the smallest dividing unit of chapter titles. The vectorization of each text fragment and target word can be performed using a pre-trained language model (e.g., Doc2Vec model, BERT model, etc.) to obtain corresponding text vectors. The first text vector can represent the semantic features of the corresponding text fragment. The second text vector can represent the semantic features of the target word.

[0060] Step S4052: Perform similarity calculations between the second text vector and the first text vector corresponding to each text segment to obtain similarity values.

[0061] The similarity value represents the degree of association between the target word and each text segment; the higher the similarity value, the stronger the association. Specific similarity calculations can include vector cosine distance, L2 distance, etc., which are not limited in this application.

[0062] Step S4053: Determine the first local text information based on the text segments corresponding to the first text vectors whose similarity values ​​satisfy the second similarity condition among multiple first text vectors.

[0063] In this embodiment, the second similarity condition can be that the similarity value reaches a certain set threshold, or it can be that the similarity values ​​are sorted from high to low and then placed within a preset number of positions. This application does not limit the specific criteria. If there are multiple first text vectors that satisfy the second similarity condition, the text segments corresponding to these multiple first text vectors can be combined from high to low similarity to obtain the first local text information. Therefore, first local text information with a high degree of relevance to the target vocabulary can be extracted from the full text information, reducing the amount of data processed by the subsequent model and improving processing efficiency.

[0064] The second prompt word template can be used to convert the target vocabulary, target intent type, and first local text information into a prompt word format suitable for input into an artificial intelligence model. For example, the second prompt word template is as follows: You will receive the following three types of information: target keywords, target intent type, and first partial text information. Based on the first partial text information, combined with the target keywords and target intent type, output the second search result. The output format must be: <result> Second search result< / result> Rules: If the target intent type is "definition," then extract and output the definition of the target term from the first local text information; if the target intent type is "case," then extract and output cases related to the target term from the first local text information; if the target intent type is "use," then extract and output the use of the target term from the first local text information; if the first local text information contains multiple related information, they need to be integrated to ensure the result is concise, coherent, and accurate; if the first local text information does not provide valid information that satisfies the intent, then... <result>Output: "Unclear"; do not add explanations, tips or any extra words in the output. The following are the search words: "+ anchor + " The following is the user's intent: "+ intent + " The following is the context content: "+ ContextSearch.

[0065] Wherein, "+ anchor +" is the placeholder corresponding to the target vocabulary, "+ intent +" is the placeholder corresponding to the target intent type, and "+ ContextSearch +" is the placeholder corresponding to the first partial text information. The second prompt word information is generated based on the target vocabulary, the target intent type, the first partial text information and the second prompt word template. Specifically, the target vocabulary, the target intent type and the first partial text information can be filled into the placeholder positions corresponding to the second prompt word template to obtain the second prompt word information. The second prompt word information can be input into an artificial intelligence model to obtain the final second search result output. The artificial intelligence model can be a large language model such as GPT4, DeepseekV3, Qwen2.5, etc., which is not limited herein. In an embodiment, after the second prompt word information is input into the artificial intelligence model, the output result of the artificial intelligence model is "unclear", and then the current display interface can be fed back with "no useful content can be searched in the document" to prompt the user to select other search links.

[0066] Step S406, input the second prompt word information into the artificial intelligence model to obtain the second search result corresponding to the second search link.

[0067] Therefore, through the second search link, the first partial text information that fits the target vocabulary can be extracted from the full text content of the target document, and the first partial text information can be further filtered and integrated by the processing capability of the artificial intelligence model to obtain the second search result that fits the target intent type.

[0068] In an embodiment, the at least one target search link determined above includes a third search link, Figure 7 A flowchart for generating a third search result based on a third search link is provided for the embodiments of the present application, as shown in Figure 7 According to the processing flow of each target search link, the target vocabulary is matched with the information search of the target intent type to obtain the search result corresponding to each target search link, specifically including the following steps: Step S407, in the case where the target search link is a third search link, performing directory recognition on the target document to obtain a directory list.

[0069] The third search link can be a search link based on local resources, and its content search scope can be a portion of the target document's content, specifically determined by directory items related to the target vocabulary. In one embodiment, since the directory is usually located at the beginning of the document, the existence of a directory page can be confirmed within the first preset number of pages of the target document. Specifically, this can be done by searching for special characters belonging to the directory page based on preset style rules to locate and extract the directory page, or by identifying the alignment features, spacing, layout style, and correlation features between each text line within the first preset number of pages to determine whether the text line belongs to the directory, and merging all text lines belonging to the directory to obtain the directory page. This application does not limit this. After determining the directory page, the directory page can be identified and split into individual directory items to obtain a directory list. This directory list can be a structured collection of directory items, specifically including information such as title text, page numbers, and hierarchical relationships.

[0070] Step S408: Generate third prompt word information based on the target vocabulary, directory list and third prompt word template, and input the third prompt word information into the artificial intelligence model to obtain the directory matching result.

[0071] The third prompt word template can be used to convert target vocabulary and directory lists into a prompt word format suitable for input into an artificial intelligence model. For example, a third prompt word template is as follows: You will receive a target vocabulary word and a list of contents. Each item in the list corresponds to a section of text. Determine in which section of text the target vocabulary word is most likely to appear and output the name of that section. Output only the name of the section, without any other explanations. The result is displayed using... <item> ...< / item> Keywords: " + anchor + " Table of Contents: " + Contents

[0072] Wherein, "+ anchor +" represents a placeholder for the target vocabulary, and "+ Contents" represents a placeholder for the directory list. Third-party suggestion information is generated based on the target vocabulary, the directory list, and the third-party suggestion template. Specifically, the target vocabulary and the directory list are filled into the placeholder positions corresponding to the third-party suggestion template to obtain the third-party suggestion information. This third-party suggestion information can be input into an artificial intelligence model to output the final third-party search result. This artificial intelligence model can be a large language model such as GPT4, DeepseekV3, or Qwen2.5; this application does not limit its scope.

[0073] Step S409: Determine the third search result corresponding to the third search link based on the target vocabulary, target intent type, target document, and directory matching results.

[0074] In one embodiment, after the third prompt word information is input into the artificial intelligence model, the directory matching result output by the artificial intelligence model does not contain " <item> ...< / item> Therefore, the current display interface can display "Unable to match a suitable directory item" to prompt the user to select another search path. In one embodiment, after the third prompt word information is input into the artificial intelligence model, the directory matching result output by the artificial intelligence model includes the directory item name, which can be compared with the directory list. If the directory item name is not recorded in the directory list, the current display interface will display "Unable to match a suitable directory item" to prompt the user to select another search path. If the directory item name is recorded in the directory list, the text content and / or context content corresponding to the directory item name can be extracted from the target document to match the target word. Since the directory matching result can provide the location range of relevant text content in the target document, the third search result corresponding to the third search path can be determined based on the target word, target intent type, target document, and directory matching result. Specifically, the relevant text content in the target document can be located with reference to the directory matching result, and the third search result corresponding to the target word can be determined from the relevant text content for the target intent type.

[0075] In one embodiment, Figure 8 A flowchart for generating a third search result based on directory matching results is provided as an embodiment of this application, such as... Figure 8 As shown, the third search result corresponding to the third search link is determined based on the target vocabulary, target intent type, target document, and directory matching results. This process includes the following steps: Step S4091: Extract the directory entry name from the directory matching result. If the directory entry name exists in the directory list, extract the second local text information of the corresponding directory entry name from the target document.

[0076] If the target item name exists in the directory list, it indicates that the prediction result of the artificial intelligence model is within the range of the input optional directory items, and the output result can be considered accurate. Accordingly, a second local text information corresponding to the target item name can be extracted from the target document. This second local text information can be the chapter text content corresponding to the target item name, or it can be the union of the chapter text content corresponding to the target item name and the adjacent context content. This application does not impose any limitations on this.

[0077] Step S4092: Generate fourth prompt information based on target vocabulary, target intent type, second local text information, and second prompt word template.

[0078] The difference between the third and second search links lies in their content search scope. The second search link extracts first local text information with a high degree of relevance to the target vocabulary from the target document, while the third search link extracts second local text information corresponding to the directory item name from the target document. Therefore, the specific process for generating the fourth prompt word information can refer to the specific process of generating the second prompt word information using the second prompt word template in the aforementioned embodiments, and will not be repeated here.

[0079] Step S4093: Input the fourth prompt word information into the artificial intelligence model to obtain the third search result corresponding to the third search link.

[0080] Therefore, the third search link can first determine the directory items that match the target words, then determine the second local text information corresponding to the directory items, and use the processing power of the artificial intelligence model to further filter and integrate the second local text information to obtain the third search result that fits the target intent type.

[0081] In one embodiment, the aforementioned identified at least one target search link includes a fourth search link. Figure 9 A flowchart for generating a fourth search result based on a fourth search link is provided for embodiments of this application, such as... Figure 9 As shown, the target keywords are matched with target intent types for information retrieval according to the processing flow corresponding to each target search link, and the search results for each target search link are obtained. The specific steps include: Step S410: When the target search link is the fourth search link, obtain the target page number range and extract the third local text information corresponding to the target page number range from the target document.

[0082] The fourth search link can be a search link based on local resources. Its content search scope can be the text content corresponding to the target page number range entered by the user. This is suitable for scenarios where the user knows the approximate location of content related to the target words. By obtaining the target page number range, the third local text information corresponding to the target page number range can be directly extracted from the target document and used to generate the subsequent fourth search result.

[0083] Step S411: Generate fifth prompt information based on target vocabulary, target intent type, third local text information, and second prompt word template.

[0084] The difference between the fourth and second search links lies in their content search scope. The second search link extracts first local text information with a high degree of relevance to the target words from the target document, while the fourth search link uses third local text information corresponding to the target page range provided by the user. Therefore, the specific process for generating the fifth prompt word information can refer to the specific process of generating the second prompt word information using the second prompt word template in the aforementioned embodiments, and will not be repeated here.

[0085] Step S412: Input the fifth prompt word information into the artificial intelligence model to obtain the fourth search result corresponding to the fourth search link.

[0086] Therefore, the fourth search link can determine the second local text information within the user-specified range, and the processing power of the artificial intelligence model can be used to further filter and integrate the second local text information to obtain the fourth search result that matches the target intent type.

[0087] Step S305: Integrate the search results corresponding to each target search link to obtain the target association information corresponding to the target words.

[0088] After obtaining the search results corresponding to each target search link, integration operations such as deduplication, sorting, or merging can be performed to make the final result more comprehensive and accurate than that of a single link. In one embodiment, for the search results corresponding to each target search link, the matching degree can be calculated separately with the target vocabulary and target intent type. The search results are then sorted from high to low according to the calculated matching degree values ​​and combined to obtain target association information. In one embodiment, the source information of the search results corresponding to each target search link can be recorded, such as a certain URL, a certain directory, or a certain page number. The search results corresponding to each target search link are merged with the source information to obtain updated search results. Then, the updated search results are combined to obtain target association information. Optionally, the target association information can be displayed on the current display interface through a preset display effect. For example, the preset display effect can be a card-style display, where each search result can be displayed separately on the current display interface as a card or note. Another example is that the preset display effect can be a table format, where the search results are filled into a preset table according to the matching degree values ​​from high to low, and the preset table is displayed on the current display interface.

[0089] In one embodiment, Figure 10 This application provides an embodiment of a diagram illustrating the selection of multiple search links and the integration of their respective search results. Figure 10 As shown in the figure, multiple search links 1001 are set up, specifically including a first search link, a second search link, a third search link, and a fourth search link. In response to the user's link selection operation, the first search link and the second search link can be determined as the target search links. Among them, the first search link outputs the first search result, and the second search link outputs the second search result. The figure integrates the first search result and the second search result in a table to obtain and display the target association information 1002 corresponding to the target words.

[0090] Since there can be one or more target search links, different processing strategies can be applied depending on the number. In one embodiment, when there is only one target search link, the search result corresponding to that target search link can be directly presented to the user. Alternatively, the system can query the stored historical data to see if there are historical search results for the same or other search links corresponding to the target term, and calculate the similarity between the historical search results and the current search result. If the similarity value exceeds a preset similarity threshold, it indicates that the two search results are related. The system can choose to directly concatenate the two search results and present them to the user, or it can deduplicate and merge the two search results before presenting them to the user. In one embodiment, Figure 11 A flowchart for integrating the search results corresponding to each target search link is provided in this application embodiment, such as... Figure 11 As shown, there are at least two target search links. The search results corresponding to each target search link are integrated to obtain the target association information corresponding to the target words. The specific steps include: Step S3051: Filter the search results corresponding to each target search link to obtain a set of similar results according to the first similarity condition.

[0091] Since search results from different target search links may overlap or be similar, semantic fusion can be used to obtain merged search results. The first similarity condition can be a quantitative rule for determining whether different search results are similar. In one embodiment, pairwise similarity values ​​can be calculated for search results from different target search links, and the first similarity condition can be that two search results are considered similar if the similarity value reaches a certain threshold. In another implementation, core keywords can be statistically analyzed for search results from different target search links according to grammatical rules, and the first similarity condition can be that the overlap rate of core keywords reaches a certain threshold. The number of similar result sets can be one or more, specifically a combination of results that meet the first similarity condition selected from various search results.

[0092] Step S3052: Semantically fuse the search results in the similar results set to obtain the merged search results.

[0093] This involves semantic fusion of search results from similar result sets. Specifically, this can be achieved by using large language models such as GPT4, DeepseekV3, and LLaMA3 to fuse content according to preset semantic alignment requirements, resulting in merged search results. These merged search results can integrate key information from the similar result sets, remove duplicate descriptions, and maintain semantic integrity.

[0094] Step S3053: Combine the merged search results with other search results that exclude similar result sets in each target search link to obtain the target association information corresponding to the target words.

[0095] Among these, the correlation between other search results is relatively low. Therefore, after obtaining the merged search results, these merged results can be combined with other search results, and the search chain corresponding to the source of each search result can be identified to ultimately obtain the target association information corresponding to the target term. Thus, in scenarios where multiple search chains are automatically executed in parallel, the output of results from multiple search chains can be simplified, and content redundancy can be reduced.

[0096] As described above, by providing users with multiple selectable search paths, users can choose one or more suitable target search paths according to their needs and characteristics, and execute the target search path selected by the user to perform information search matching the target intent type, thereby obtaining comprehensive and accurate target-related information.

[0097] Figure 12 This is a structural block diagram of a text search device provided in an embodiment of this application. The device is configured to execute the text search method provided in the above embodiment, and has corresponding functional modules and beneficial effects for executing the method. For example... Figure 12 As shown, the device specifically includes: The text information determination module 1201 is configured to determine target text information from the currently displayed page of the target document in response to a text selection operation; The vocabulary intent determination module 1202 is configured to identify and determine target words and corresponding target intent types based on target text information. The association information determination module 1203 is configured to perform information search based on at least one of the multiple search links set, the target word and the corresponding target intent type, to obtain the target association information corresponding to the target word.

[0098] As described above, by responding to the user's text selection, the system can accurately extract text information of interest to the user while eliminating interference from irrelevant content in the document. Based on the recognition of the target text information, the system can extract the most crucial target words and determine the target intent type, making subsequent search results more aligned with the user's actual needs. By using a set search path based on target words and target intent type, the system accurately obtains related information about the target words, satisfying diverse search needs. This solution combines the user's selected text with the recognition of the user's actual intent, and uses a set search path to perform text searches matching that intent, providing accurate search results that fit the user's actual query needs and improve the user experience.

[0099] In one possible embodiment, the number of target words is at least two, and the target intent types corresponding to at least two target words are the same; the association information determination module 1203 is further configured to: Information is searched based on at least one or two target words and their corresponding target intent types from multiple search links to obtain integrated target association information corresponding to at least two target words.

[0100] In one possible embodiment, the number of target words is at least two, and the target intent types corresponding to at least two target words are partially or completely different; the association information determination module 1203 is further configured to: Information is searched based on at least one or two target words and their corresponding target intent types from multiple search links to obtain integrated target association information corresponding to at least two target words.

[0101] In one possible embodiment, the association information determination module 1203 is further configured to: In response to the link selection operation, at least one target search link is determined from the set number of search links; For each target word, the following processing is performed: Information on matching target intent types is searched for each target word according to the processing flow corresponding to each target search link, resulting in search results for each target search link; the search results corresponding to each target search link are then integrated to obtain the target association information corresponding to the target word. The target association information corresponding to each target word is integrated to obtain integrated target association information.

[0102] In one possible embodiment, the association information determination module 1203 is further configured to: In response to the link selection operation, at least one target search link is determined from the set number of search links; Based on the processing flow corresponding to each target search link, the target words are matched with the target intent type information search to obtain the search results corresponding to each target search link; The search results corresponding to each target search link are integrated to obtain the target association information corresponding to the target words.

[0103] In one possible embodiment, the at least one target search link includes a first search link, and the association information determination module 1203 is further configured to: When the target search path is the first search path, relevant content information is obtained by performing network retrieval based on the target keywords; First prompt word information is generated based on target vocabulary, target intent type, relevant content information, and first prompt word template; The first prompt word information is input into the artificial intelligence model to obtain the first search result corresponding to the first search link.

[0104] In one possible embodiment, the number of target search links is at least two, and the association information determination module 1203 is further configured to: The search results corresponding to each target search link are filtered into a similar result set according to the first similarity condition; The search results in the similar results set are semantically fused to obtain the merged search results; The merged search results and other search results that exclude similar result sets in each target search link are combined to obtain the target association information corresponding to the target words.

[0105] In one possible embodiment, the at least one target search link includes a second search link, and the association information determination module 1203 is further configured to: When the target search path is the second search path, the target document is parsed to determine the full text information; Based on the target vocabulary, first local text information that meets the second similarity condition is determined from the full text information; second prompt word information is generated based on the target vocabulary, target intent type, first local text information and second prompt word template. The second prompt word information is input into the artificial intelligence model to obtain the second search result corresponding to the second search link.

[0106] In one possible embodiment, the association information determination module 1203 is further configured to: The entire text information is segmented into multiple text fragments. Each text fragment is vectorized to obtain the first text vector, and the target words are vectorized to obtain the second text vector. The similarity value is obtained by comparing the second text vector with the first text vector corresponding to each text segment; The first local text information is determined based on the text segments corresponding to the first text vectors whose similarity values ​​satisfy the second similarity condition among multiple first text vectors.

[0107] In one possible embodiment, the at least one target search link includes a third search link, and the association information determination module 1203 is further configured to: When the target search link is a third search link, the target document is identified to obtain a directory list; Third-party prompt information is generated based on the target vocabulary, the directory list, and the third-party prompt word template. The third-party prompt word information is then input into an artificial intelligence model to obtain the directory matching results. The third search result corresponding to the third search link is determined based on the target vocabulary, target intent type, target document, and directory matching results.

[0108] In one possible embodiment, the association information determination module 1203 is further configured to: Extract directory entry names from the directory matching results. If a directory entry name exists in the directory list, extract the second local text information of the corresponding directory entry name from the target document. Generate fourth prompt information based on target vocabulary, target intent type, second local text information, and second prompt word template; The fourth prompt word information is input into the artificial intelligence model to obtain the third search result corresponding to the third search link.

[0109] In one possible embodiment, the at least one target search link includes a fourth search link, and the association information determination module 1203 is further configured to: When the target search link is the fourth search link, obtain the target page number range and extract the third local text information corresponding to the target page number range from the target document; The fifth prompt word information is generated based on the target vocabulary, target intent type, third local text information, and second prompt word template; The fifth prompt word information is input into the artificial intelligence model to obtain the fourth search result corresponding to the fourth search link.

[0110] In one possible embodiment, the text information determination module 1201 is further configured to: The text information of the currently displayed page is determined by parsing the document type of the target document. Based on the coordinate information of the region corresponding to the text selection operation, the target text information is determined from the page text information.

[0111] In one possible embodiment, the vocabulary intent determination module 1202 is further configured to: The sixth prompt word information is generated based on the target text information and the fourth prompt word template. The sixth prompt word information is then input into the artificial intelligence model to obtain at least one candidate word and the candidate intent type corresponding to each candidate word. In response to the information selection operation, a target word and its corresponding target intent type are determined from at least one candidate word and the candidate intent type corresponding to each candidate word.

[0112] Figure 13 This is a schematic diagram of the structure of a text search device provided in an embodiment of this application, as shown below. Figure 13 As shown, the device includes a processor 1301, a memory 1302, an input device 1303, and an output device 1304; the number of processors 1301 in the device can be one or more. Figure 13 Taking a processor 1301 as an example; the processor 1301, memory 1302, input device 1303, and output device 1304 in the device can be connected via a bus or other means. Figure 13 Taking a bus connection as an example, the memory 1302, as a computer-readable storage medium, can be configured to store software programs, computer-executable programs, and modules, such as the program instructions / modules corresponding to the text search method in this embodiment. The processor 1301 executes various functional applications and data processing of the device by running the software programs, instructions, and modules stored in the memory 1302, thereby implementing the aforementioned text search method. The input device 1303 can be configured to receive input numeric or character information and generate key signal inputs related to user settings and function control of the device. The output device 1304 may include a display screen or other display device.

[0113] The text search device provided above can be used to execute the text search method provided in any of the above embodiments, and has corresponding functions and beneficial effects.

[0114] This application also provides a non-volatile storage medium containing computer-executable instructions, which, when executed by a computer processor, are configured to perform a text search method described in the above embodiments, comprising: in response to a text selection operation, determining target text information from the currently displayed page of a target document; identifying target words and corresponding target intent types based on the target text information; and performing information search based on at least one of a plurality of search links, the target words, and the corresponding target intent types to obtain target association information corresponding to the target words.

[0115] Storage medium – any type of memory device or storage device. The term "storage medium" is intended to include: mounting media, such as CD-ROM, floppy disk, or magnetic tape devices; computer system memory or random access memory, such as DRAM, DDR RAM, SRAM, EDO RAM, Rambus RAM, etc.; non-volatile memory, such as flash memory, magnetic media, optical storage; registers or other similar types of memory elements, etc. Storage medium may also include other types of memory or combinations thereof. Furthermore, storage medium may reside in a first computer system in which the program is executed, or it may reside in a different second computer system connected to the first computer system via a network (such as the Internet). The second computer system can provide program instructions to the first computer for execution. The term "storage medium" may include two or more storage media residing in different locations (e.g., in different computer systems connected via a network). Storage medium may store program instructions (e.g., specifically implemented as a computer program) executable by one or more processors.

[0116] Of course, the computer-executable instructions provided in the embodiments of this application are not limited to the text search method described above, but can also perform related operations in the text search method provided in any embodiment of this application.

[0117] It should be noted that the numbering of each step in this solution is only used to describe the overall design framework of this solution and does not indicate a necessary sequential relationship between the steps. As long as the overall implementation process conforms to the overall design framework of this solution, it falls within the protection scope of this solution. The literal order in the description is not an exclusive limitation on the specific implementation process of this solution. Those skilled in the art should understand that the embodiments of this application can be provided as methods, systems, or computer program products. In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory. Memory may include non-persistent memory in computer-readable media, random access memory (RAM), and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0118] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0119] Note that the above description is merely a preferred embodiment of the present invention and the technical principles employed. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of the present invention, the scope of which is determined by the scope of the appended claims.< / result> < / result>

Claims

1. A text search method, characterized in that, include: In response to a text selection operation, determine the target text information from the currently displayed page of the target document; Based on the target text information, target words and corresponding target intent types are identified and determined; Information search is performed based on at least one of the multiple search links set, the target vocabulary and the corresponding target intent type to obtain the target association information corresponding to the target vocabulary.

2. The text search method according to claim 1, characterized in that, The number of target words is at least two, and the target intent types corresponding to at least two target words are the same; The information search is performed based on at least one of the multiple search links set, the target vocabulary and the corresponding target intent type to obtain the target association information corresponding to the target vocabulary, including: Information is searched based on at least one or at least two of the target words and their corresponding target intent types from multiple search links to obtain integrated target association information corresponding to at least two of the target words.

3. The text search method according to claim 1, characterized in that, The number of target words is at least two, and the target intent types corresponding to at least two target words are partially or completely different; The information search is performed based on at least one of the multiple search links set, the target vocabulary and the corresponding target intent type to obtain the target association information corresponding to the target vocabulary, including: Information is searched based on at least one or at least two of the target words and their corresponding target intent types from multiple search links to obtain integrated target association information corresponding to at least two of the target words.

4. The text search method according to claim 2 or 3, characterized in that, The information search is performed based on at least one or at least two of the target words and their corresponding target intent types from multiple set search links to obtain integrated target association information corresponding to at least two of the target words, including: In response to the link selection operation, at least one target search link is determined from the set multiple search links; For each target word, the following processing is performed: Information search matching the target intent type is performed on the target word according to the processing flow corresponding to each target search link to obtain search results for each target search link; the search results corresponding to each target search link are integrated to obtain target association information corresponding to the target word. The target association information corresponding to each target word is integrated to obtain the integrated target association information.

5. The text search method according to claim 1, characterized in that, The information search is performed based on at least one of the multiple search links set, the target vocabulary, and the corresponding target intent type to obtain the target association information corresponding to the target vocabulary, including: In response to the link selection operation, at least one target search link is determined from the set number of search links; According to the processing flow corresponding to each target search link, the target words are matched with the target intent type to obtain the search results corresponding to each target search link; The search results corresponding to each of the target search links are integrated to obtain the target association information corresponding to the target words.

6. The text search method according to claim 5, characterized in that, The at least one target search link includes a first search link; The step of performing information search on the target words according to the target intent type based on the processing flow corresponding to each target search link, to obtain the search results corresponding to each target search link, includes: When the target search link is the first search link, relevant content information is obtained by performing a network search based on the target vocabulary. First prompt word information is generated based on the target vocabulary, the target intent type, the relevant content information, and the first prompt word template; The first prompt word information is input into the artificial intelligence model to obtain the first search result corresponding to the first search link.

7. The text search method according to claim 5, characterized in that, The number of target search links is at least two, and the process of integrating the search results corresponding to each target search link to obtain the target association information corresponding to the target term includes: The search results corresponding to each of the target search links are filtered into a set of similar results according to the first similarity condition; The search results in the similar result set are semantically fused to obtain the merged search results; The merged search results and other search results excluding the similar result set in each of the target search links are combined to obtain the target association information corresponding to the target vocabulary.

8. The text search method according to claim 5, characterized in that, The at least one target search link includes a second search link; The step of performing information search on the target words according to the target intent type based on the processing flow corresponding to each target search link, to obtain the search results corresponding to each target search link, includes: When the target search link is the second search link, the target document is parsed to determine the full text information; Based on the target vocabulary, first local text information that satisfies the second similarity condition is determined from the full text information, and second prompt word information is generated based on the target vocabulary, the target intent type, the first local text information and the second prompt word template. The second prompt word information is input into the artificial intelligence model to obtain the second search result corresponding to the second search link.

9. The text search method according to claim 8, characterized in that, The step of determining the first local text information that satisfies the second similarity condition from the full text information based on the target vocabulary includes: The full text information is segmented into multiple text segments, each text segment is vectorized to obtain a first text vector, and the target words are vectorized to obtain a second text vector. The similarity value is obtained by performing similarity calculations between the second text vector and the first text vector corresponding to each of the text segments. The first local text information is determined based on the text segments corresponding to the first text vectors whose similarity values ​​satisfy the second similarity condition among multiple first text vectors.

10. The text search method according to claim 5, characterized in that, The at least one target search link includes a third search link; The step of performing information search on the target words according to the target intent type based on the processing flow corresponding to each target search link, to obtain the search results corresponding to each target search link, includes: When the target search link is a third search link, the target document is identified to obtain a directory list; Based on the target vocabulary, the directory list, and the third prompt word template, third prompt word information is generated, and the third prompt word information is input into an artificial intelligence model to obtain the directory matching result; The third search result corresponding to the third search link is determined based on the target vocabulary, the target intent type, the target document, and the directory matching result.

11. The text search method according to claim 10, characterized in that, The step of determining the third search result corresponding to the third search link based on the target vocabulary, the target intent type, the target document, and the directory matching result includes: Extract the directory entry name from the directory matching results, and if the directory entry name exists in the directory list, extract the second local text information corresponding to the directory entry name from the target document; A fourth prompt word is generated based on the target vocabulary, the target intent type, the second local text information, and the second prompt word template; The fourth prompt word information is input into the artificial intelligence model to obtain the third search result corresponding to the third search link.

12. The text search method according to claim 5, characterized in that, The at least one target search link includes a fourth search link; The step of performing information search on the target words according to the target intent type based on the processing flow corresponding to each target search link, to obtain the search results corresponding to each target search link, includes: When the target search link is the fourth search link, the target page number range is obtained, and the third local text information corresponding to the target page number range is extracted from the target document; The fifth prompt word information is generated based on the target vocabulary, the target intent type, the third local text information, and the second prompt word template; The fifth prompt word information is input into the artificial intelligence model to obtain the fourth search result corresponding to the fourth search link.

13. The text search method according to any one of claims 5-12, characterized in that, The step of determining the target text information from the currently displayed page of the target document in response to a text selection operation includes: The currently displayed page is parsed and the page text information is determined based on the document type of the target document; Based on the region coordinate information corresponding to the text selection operation, the target text information is determined from the page text information.

14. The text search method according to any one of claims 5-12, characterized in that, The process of identifying and determining target words and corresponding target intent types based on the target text information includes: Based on the target text information and the fourth prompt word template, a sixth prompt word information is generated, and the sixth prompt word information is input into an artificial intelligence model to obtain at least one candidate word and a candidate intent type corresponding to each candidate word; In response to the information selection operation, a target word and a corresponding target intent type are determined from the at least one candidate word and the candidate intent type corresponding to each candidate word.

15. A text search device, characterized in that, include: The text information determination module is configured to determine the target text information from the currently displayed page of the target document in response to a text selection operation. The vocabulary intent determination module is configured to identify and determine target words and corresponding target intent types based on the target text information. The association information determination module is configured to perform information search based on at least one of a plurality of set search links, the target word and the corresponding target intent type, to obtain the target association information corresponding to the target word.

Images