Software system-based search method and system

Abstract

The invention discloses a search method and system based on a software system, and the method comprises the steps: constructing an index, and converting a query request inputted by a current user into a preset query expression; performing search query of the current user in the constructed index, performing comprehensive evaluation on a key index, a timeliness index, an authority index and a downloading index of each group of documents in the screened document set after the search query is executed, and measuring the correlation between the documents and the user query from multiple angles; the key index considers the matching condition of keywords at different positions of the document, and the timeliness index reflects the timeliness of the document; the authority index is combined with reference times and user feedback evaluation; the downloading index considers the network speed of a user and the size of a document; the multi-dimensional evaluation enables the search result to more accurately reflect the user demand, and the correlation is higher.

Description

Technical Field

[0001] The present invention relates to the technical field of data search, and in particular to a search method and system based on a software system. Background Art

[0002] With the rapid development of information technology, the amount of data is growing explosively. Various types of data, including text, pictures, videos, audio, etc., are widely present in different sources such as databases, file systems, and networks. When faced with massive amounts of data, users need a search system that can quickly and accurately find the information they need.

[0003] However, the software system-based search method in the prior art still has the following deficiencies:

[0004] In the search process, the relevance of documents to user queries is often measured from a single perspective, without multi-dimensional evaluation of documents and user queries from multiple perspectives, such as authority, etc., resulting in low comprehensiveness and accuracy of the search;

[0005] At the same time, when conducting multi-dimensional evaluations of documents, personalized recommendations are not made based on user needs, and the level of intelligence is low.

[0006] Therefore, a search method and system based on software system are introduced. Summary of the invention

[0007] In view of this, the present invention provides a search method and system based on a software system to solve the problems raised by the above background technology.

[0008] The object of the present invention can be achieved by the following technical solution: A search method based on a software system, comprising:

[0009] Building an index: preprocess the collected data and build an index structure based on the preprocessed data;

[0010] Processing queries: converting the query request entered by the current user into a preset query expression; and executing the current user's search query in the constructed index;

[0011] Search matching: Use the constructed index to locate each group of data that matches the current user's query expression, and execute the corresponding steps to obtain the filtered document set, parse and evaluate each group of documents in the filtered document set, so as to obtain the relevant evaluation index Gr of each group of documents in the current user's filtered document set, where r represents the number of each group of documents in the filtered document set, r = 1, 2, ..., k, k represents the total number of documents in the filtered document set, and sort based on the size of the relevant evaluation index Gr of each group of documents in the filtered document set;

[0012] Output result: After sorting based on the size of the relevant evaluation index Gr of each group of documents in the filtered document collection, the sorted search results are displayed in the format preset by the current user and pushed to the current user. After receiving the pushed content, the current user can selectively perform the corresponding steps to select the final document for viewing.

[0013] In some embodiments, corresponding steps are performed to obtain a filtered document set, specifically:

[0014] S1: Extract each keyword in the current user query expression according to the constructed index structure, locate the document containing each keyword based on each extracted keyword, integrate the located groups of documents, and thus construct a document set corresponding to the current user query expression;

[0015] S2: After building the document collection, parse the current user query expression and input the pre-built logical operator set for matching; the logical operators include "and", "or" and "not";

[0016] S3: If the match is successful, the logical operator of the successful match is parsed. If the parsing result is "and", the documents that meet each keyword at the same time are filtered out from the document collection to build a filtered document collection. If the parsing result is "or", the constructed document collection is directly used as the filtered document collection; if the parsing result is "no", the keyword corresponding to "no" is first identified. After the identification is completed, the documents that do not contain negative keywords are filtered out from the document collection to build a filtered document collection.

[0017] In some embodiments, each set of documents in the filtered document set is parsed and evaluated, specifically:

[0018] M1: Parse and filter the content of each group of documents in the document collection, remove negative keywords from each keyword in the current user query expression, match it with the content of each group of documents, count the number of keywords appearing in the content of each group of documents, divide the content of each group of documents into corresponding positions according to the title, abstract and content, and classify the number of keywords appearing in the content of each group of documents according to the divided positions, so as to obtain the number of keywords appearing in different divided positions of the content of each group of documents; set a weight coefficient corresponding to each divided position;

[0019] The number of keywords appearing in different division positions of each group of document content is multiplied by the corresponding weight coefficient, and then summed up to obtain the criticality index e of each group of documents in the filtered document set;

[0020] M2: Mark the time point when the current user inputs the query request as the starting point, preset three groups of time intervals, integrate the starting point with the preset three groups of time intervals, and obtain three groups of time windows for the current user, which are divided into recent time windows, medium-term time windows, and early time windows; obtain the release time of each group of documents in the filtered document collection, and match them with the three groups of time windows, set each group of time windows to correspond to a reference score, and use the reference score corresponding to each group of documents as the timeliness index r of each group of documents.

[0021] In some embodiments, parsing and evaluating each set of documents in the filtered document set further includes:

[0022] M3: Count the number of citations of each set of document content, and use the preset conversion rules to convert the number of citations of each set of document content into value scores;

[0023] Preset the range of values ​​for each group of citations corresponding to the number of citations, and set each group of citations to correspond to a value score; match the number of citations of each group of document content with the range of values ​​for each group of citations, so as to obtain the value score corresponding to each group of document content;

[0024] Obtain the number of user feedback evaluations for each group of document content, extract the score corresponding to each user feedback evaluation from the number of user feedback evaluations for each group of document content, and take the average to obtain the average user score for each group of document content;

[0025] The weight coefficients corresponding to the preset value scores and user score averages are respectively multiplied by the corresponding weight coefficients, and then the sum is calculated to obtain the authority index p of each group of document content.

[0026] In some embodiments, parsing and evaluating each set of documents in the filtered document set further includes:

[0027] M4: Send a network speed test data packet to the current user. The current user downloads after receiving the network speed test data packet. Record the time from the start of downloading to obtain the download speed value of the current user at each time point within the set time period. Use the standard deviation formula to calculate the download speed value of the current user at each time point within the set time period to obtain the network speed fluctuation value. Set a reference threshold for the network speed fluctuation value. If the network speed fluctuation value of the current user is greater than the set reference threshold, take the median value from the download speed values ​​at each time point as the network speed evaluation value. Otherwise, take the mean of the download speed values ​​at each time point as the network speed evaluation value.

[0028] Based on the network speed evaluation value of the current user, the network speed of the current user is graded into low speed, medium speed and high speed, and a maximum value of the allowed download file is set for low speed, medium speed and high speed respectively;

[0029] Set low speed, medium speed and high speed to correspond to a network speed range respectively, match the current user's network speed evaluation value with the corresponding network speed range, so as to complete the classification of user network speed;

[0030] The file size of each group of document contents is obtained, and the ratio is calculated with the maximum value of the allowable downloadable file corresponding to the current user network speed evaluation value, so as to obtain the downloadability index m of each group of documents.

[0031] In some embodiments, the relevant evaluation index Gr of each group of documents in the current user filtered document set is obtained, specifically:

[0032] M5: Extract the criticality index e, timeliness index r, authority index p and downloadability index m corresponding to each set of document content, according to the formula where e 期望 、r 期望 and p 期望 They respectively represent the minimum expected criticality index, minimum expected timeliness index and minimum expected authority index pre-set by the current user; among them, a1, a2, a3 and a4 respectively represent the impact weight factors corresponding to the criticality index e, timeliness index r, authority index p and downloadability index m.

[0033] In some embodiments, the corresponding steps are selectively performed to select the final document, specifically:

[0034] Selection bias: After receiving the pushed content, if the user is not satisfied with the currently pushed sorted search results, the adjustment signaling is triggered and a selection interface is pushed to the current user. The selection interface includes key bias screening, timeliness bias screening, authority bias screening and download bias screening. After the current user selects the corresponding bias screening through the selection interface, the values ​​of the influence weight factors a1, a2, a3 and a4 of the key index e, timeliness index r, authority index p and download index m corresponding to each group of document content are adjusted according to the preset of the corresponding bias screening;

[0035] Reselect: If the user is still not satisfied with the content currently pushed through the push selection interface, then after removing the negative keywords from each keyword in the current user query expression, synonym matching and conversion are performed, and steps S1-S3 and steps M1-M5 are executed again.

[0036] A search system based on a software system, comprising:

[0037] Construction module: build index structure for collected data;

[0038] Processing module: converts the query request entered by the current user into a preset query expression, and executes the current user's search query in the constructed index;

[0039] Matching module: locates each group of data matching the current user query expression, and executes corresponding steps to obtain a filtered document set, and parses and evaluates each group of documents in the filtered document set through a pre-established relevance scoring model, thereby obtaining a relevance evaluation index Gr of each group of documents in the current user filtered document set, where r represents the number of each group of documents in the filtered document set, r = 1, 2, ..., k, k represents the total number of documents in the filtered document set, and sorts the documents based on the relevance evaluation index Gr of each group of documents in the filtered document set;

[0040] Output module: Display the sorted search results in the format preset by the current user and push them to the current user. After receiving the pushed content, the current user can selectively perform corresponding steps to select the final document for viewing.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] The present invention constructs an index and converts the query request input by the current user into a preset query expression; executes the current user's search query in the constructed index, and after execution, comprehensively evaluates the key index, timeliness index, authority index and downloadability index of each group of documents in the screened document set, and measures the relevance of the document to the user query from multiple perspectives; the key index considers the keyword matching situation at different positions of the document, the timeliness index reflects the timeliness of the document; the authority index combines the number of citations and user feedback evaluation; the downloadability index considers the user's network speed and the size of the document; this multi-dimensional evaluation enables the search results to more accurately reflect user needs and have higher relevance;

[0043] The present invention uses the constructed index to locate each group of data that matches the current user query expression, parses the current user query expression and inputs a pre-constructed set of logical operators for matching, and executes corresponding steps based on the matching results to obtain a filtered document set, so that the search results are more orderly and organized, which is convenient for users to quickly find the required information and reduces the time for users to filter useful information from a large number of irrelevant results;

[0044] When a user is not satisfied with the pushed sorted search results, the present invention can adjust the influence weight factors of the key index, timeliness index, authority index and downloadability index corresponding to each group of document contents by selecting biased filtering, including key biased filtering, timeliness biased filtering, authority biased filtering and download biased filtering, so that the search results are more in line with user needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Further details, features and advantages of the present application are disclosed in the following description of exemplary embodiments in conjunction with the accompanying drawings, in which:

[0046] Figure 1 is a flow chart of the present invention;

[0047] Figure 2 It is a principle block diagram of the present invention. DETAILED DESCRIPTION

[0048] Several embodiments of the present application will be described in more detail below with reference to the accompanying drawings so that those skilled in the art can implement the present application. The present application can be embodied in many different forms and purposes and should not be limited to the embodiments described herein. These embodiments are provided to make the present application comprehensive and complete, and to fully convey the scope of the present application to those skilled in the art. The embodiments do not limit the present application.

[0049] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those of ordinary skill in the art to which this application belongs. It will be further understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant art and / or this specification, and will not be interpreted in an idealized or overly formal sense, unless explicitly defined as such herein.

[0050] Example 1

[0051] See also Figure 1 As shown, a search method based on a software system includes:

[0052] Building an index: preprocessing the collected data; preprocessing includes data cleaning, data standardization, stop word removal, and stem extraction; building an index structure for the preprocessed data; the index structure is specifically an inverted index;

[0053] Processing queries: By constructing a parsing window in the software system, the parsing window includes basic components such as a text input box and a search button, and using the constructed parsing window to convert the query request entered by the current user into a preset query expression; for example, converting the natural language query entered by the user into a query expression composed of keywords, logical operators, etc.; and executing the current user's search query in the constructed index;

[0054] Search matching: Use the constructed index to locate each group of data that matches the current user's query expression, and execute the corresponding steps to obtain the filtered document set, parse and evaluate each group of documents in the filtered document set, so as to obtain the relevant evaluation index Gr of each group of documents in the current user's filtered document set, where r represents the number of each group of documents in the filtered document set, r = 1, 2, ..., k, k represents the total number of documents in the filtered document set, and sort based on the size of the relevant evaluation index Gr of each group of documents in the filtered document set;

[0055] Execute the corresponding steps to obtain the filtered document collection, specifically:

[0056] S1: Extract each keyword in the current user query expression according to the constructed index structure, locate the document containing each keyword based on each extracted keyword, integrate the located groups of documents, and thus construct a document set corresponding to the current user query expression;

[0057] S2: After building the document collection, parse the current user query expression and input the pre-built logical operator set for matching; the logical operators include "and", "or" and "not";

[0058] S3: If the match is successful, the logical operator of the successful match is parsed. If the parsing result is "and", the documents that meet each keyword at the same time are filtered out from the document collection to build a filtered document collection. If the parsing result is "or", the constructed document collection is directly used as the filtered document collection. If the parsing result is "no", the keyword corresponding to "no" is first identified. After the identification is completed, the documents that do not contain negative keywords are filtered out from the document collection to build a filtered document collection.

[0059] For example, there is a document database that contains multiple documents on technology, medical care, entertainment, etc., and an inverted index has been built. The user enters the query expression "(artificial intelligence or machine learning) and medical applications and not image recognition"

[0060] Extracting keywords “artificial intelligence”, “machine learning”, “medical application”, and “image recognition” from the query expression;

[0061] Through the inverted index, find the document set containing "artificial intelligence", assuming it is {document 1, document 2, document 3, document 5, document 7}; the document set containing "machine learning", assuming it is {document 2, document 4, document 6, document 7, document 8}; the document set containing "medical application", assuming it is {document 3, document 5, document 7, document 9, document 10}; the document set containing "image recognition", assuming it is {document 2, document 5, document 7};

[0062] Integrate these located documents together to obtain a large document set {document 1, document 2, document 3, document 4, document 5, document 6, document 7, document 8, document 9, document 10};

[0063] Parse the query expression "(artificial intelligence or machine learning) and medical application and not image recognition" input by the user and match it with a set of logical operators ("and", "or", "no");

[0064] First, when we encounter "(artificial intelligence or machine learning)", we do not filter it because it is an "or" relationship, and keep the previously integrated document set {Document 1, Document 2, Document 3, Document 4, Document 5, Document 6, Document 7, Document 8, Document 9, Document 10};

[0065] Processing the "and" relationship: Next, we encounter "(artificial intelligence or machine learning) and medical applications", which is an "and" relationship. Therefore, we need to filter out documents that meet both "(artificial intelligence or machine learning)" and "medical applications" from the current document collection; by checking the document content, the filtered document collection becomes {Document 3, Document 5, Document 7};

[0066] Processing "no" relationship: Finally, we encounter "and not image recognition", which is a "no" relationship. We identify the negative keyword "image recognition", and then filter out the documents that do not contain "image recognition" from the current document set {document 3, document 5, document 7}, and finally get the filtered document set {document 3};

[0067] Each set of documents in the filtered document collection is parsed and evaluated, specifically:

[0068] M1: Parse and filter the content of each group of documents in the document collection, remove negative keywords from each keyword in the current user query expression, match it with the content of each group of documents, count the number of keywords appearing in the content of each group of documents, divide the content of each group of documents into corresponding positions according to the title, abstract and content, and classify the number of keywords appearing in the content of each group of documents according to the divided positions, so as to obtain the number of keywords appearing in different divided positions of the content of each group of documents; set a weight coefficient corresponding to each divided position;

[0069] The number of keywords appearing in different division positions of each group of document content is multiplied by the corresponding weight coefficient, and then summed up to obtain the criticality index e of each group of documents in the filtered document set;

[0070] It should be noted that the calculation of the criticality index e provides a more scientific quantitative standard for the ranking of search results; it comprehensively considers the occurrence of keywords in different key positions of the document, so that documents with high relevance are ranked first;

[0071] Since the keywords in different parts of the document are comprehensively considered, a certain type of document will not be overly concentrated in the top of the search results simply because there are many keyword matches in the main text; documents with different structural characteristics have the opportunity to be reasonably displayed according to their overall relevance scores, making the search results more diverse and comprehensive, and meeting users' needs for information richness.

[0072] M2: Take the time point when the current user inputs the query request as the starting point, preset three groups of time intervals, integrate the starting point with the preset three groups of time intervals, and obtain three groups of time windows for the current user, which are divided into recent time windows, medium-term time windows, and early time windows; for example, the recent time window is set to within 30 days from the starting time point, the medium-term time window is 31-180 days, and the early time window is 181 days and before; obtain the release time of each group of documents in the filtered document collection, and match them with the three groups of time windows, set each group of time windows to correspond to a reference score, and use the reference score corresponding to each group of documents as the timeliness index r of each group of documents; the timeliness score range is set to 1-5 and is an integer;

[0073] It should be noted that the reference scores of the three groups of time windows are compared according to their size, specifically: recent time window > mid-term time window > early time window;

[0074] The calculated timeliness index r gives each document a quantitative indicator reflecting its timeliness, which can be used for subsequent search result evaluation and sorting operations. In this way, search results can be more inclined to display documents with strong timeliness, meeting users' needs for timely information.

[0075] M3: Count the number of citations of each set of document content, and use the preset conversion rules to convert the number of citations of each set of document content into value scores;

[0076] Citation counts can assess the impact and importance of a document;

[0077] The preset reference counts correspond to the value ranges of each group of times, and each group of times value ranges is set to correspond to a value score; the value score is set to 1-5 and is a positive integer. The more references there are, the higher the corresponding value score will be; the reference counts of each group of document content are matched with the value ranges of each group of times, so as to obtain the value score corresponding to each group of document content;

[0078] The number of user feedback evaluations for each set of document content is obtained; the user feedback evaluation adopts a scoring system, and the range is set to 1-5, and the scoring result is a positive integer; the score corresponding to each user feedback evaluation is extracted from the number of user feedback evaluations for each set of document content, and the average is taken to obtain the average user score of each set of document content;

[0079] Preset the weight coefficients corresponding to the value score and the average user score, multiply the value score and the average user score of each group of document content at the current time point by the corresponding weight coefficient, and then sum them up to obtain the authority index p of each group of document content;

[0080] It should be noted that incorporating the authority index into the search result ranking makes it more likely that documents with high authority will be ranked at the top; this helps users find documents more quickly that are not only relevant to the query in terms of content, but also have guaranteed quality and influence.

[0081] M4: Send a network speed test data packet to the current user. The current user downloads after receiving the network speed test data packet. Record the time from the start of downloading to obtain the download speed value of the current user at each time point within the set time period. Use the standard deviation formula to calculate the download speed value of the current user at each time point within the set time period to obtain the network speed fluctuation value. Set a reference threshold for the network speed fluctuation value. If the network speed fluctuation value of the current user is greater than the set reference threshold, take the median value from the download speed values ​​at each time point as the network speed evaluation value. Otherwise, take the mean of the download speed values ​​at each time point as the network speed evaluation value.

[0082] Based on the current user's network speed evaluation value, the current user's network speed is graded into low speed, medium speed and high speed, and a maximum value of the allowed download file is set for low speed, medium speed and high speed respectively; it is preset by the technical staff and can be adjusted later;

[0083] Set low speed, medium speed and high speed to correspond to a network speed range respectively, match the current user's network speed evaluation value with the corresponding network speed range, so as to complete the classification of user network speed;

[0084] Obtain the file size of each group of document contents, and calculate the ratio between the file size and the maximum allowable download file size corresponding to the current user network speed evaluation value, so as to obtain the downloadability index m of each group of documents;

[0085] It should be noted that, through the calculated downloadability index m, if the downloadability index m of a group of document contents is larger, it means that the size of the group of documents is close to or larger than the download size allowed by the user's network speed. When sorting, the group of document contents will be ranked later to avoid the subsequent users' downloading time of too long affecting the user experience, and further optimize the search sorting;

[0086] M5: Extract the criticality index e, timeliness index r, authority index p and downloadability index m corresponding to each set of document content, according to the formula where e 期望 、r 期望 and p 期望 They represent the minimum expected criticality index, the minimum expected timeliness index and the minimum expected authority index preset by the current user respectively; a1, a2, a3 and a4 represent the impact weight factors corresponding to the criticality index e, the timeliness index r, the authority index p and the downloadability index m respectively;

[0087] It should be noted that the formula allows users to pre-set the minimum expected criticality index, timeliness index and authority index. Users can adjust the screening criteria of search results according to their own needs and preferences. For example, professional users may have higher requirements for authority and can increase the corresponding expected index to make the search results more in line with personalized usage scenarios.

[0088] Through the above series of evaluation and calculation methods, a large number of documents can be screened and sorted more efficiently. This method based on multi-indicator comprehensive evaluation can more quickly determine the degree of match between documents and user queries. When processing large-scale document databases, it can effectively reduce the consumption of computing resources and improve the response speed and performance of the search system.

[0089] Output results: After sorting based on the size of the relevant evaluation index Gr of each group of documents in the filtered document collection, the sorted search results are displayed in the format preset by the current user and pushed to the current user. After receiving the pushed content, the current user can selectively perform corresponding steps to select the final document for viewing;

[0090] Selectively perform the corresponding steps to select the final document, specifically:

[0091] Selection bias: After receiving the pushed content, if the user is not satisfied with the currently pushed sorted search results, the adjustment signaling is triggered and a selection interface is pushed to the current user. The selection interface includes key bias screening, timeliness bias screening, authority bias screening and download bias screening. After the current user selects the corresponding bias screening through the selection interface, the values ​​of the influence weight factors a1, a2, a3 and a4 of the key index e, timeliness index r, authority index p and download index m corresponding to each group of document content are adjusted according to the preset of the corresponding bias screening;

[0092] It should be noted that if the corresponding biased screening is selected, the corresponding impact weight factor value will be relatively large.

[0093] Reselect: If the user is still not satisfied with the content currently pushed through the push selection interface, then each keyword in the current user query expression is removed from the negative keywords, and then synonyms are matched and converted; by calculating the cosine similarity of the word vectors, other words that are semantically similar to the original keywords are found; and steps S1-S3 and steps M1-M5 are executed again;

[0094] It should be noted that by matching and converting keywords into synonyms, the search is no longer limited to the keywords initially entered by the user, but covers semantically similar words, which can mine more documents that may be related to user needs and avoid missing out on valuable information due to specific words selected by the user.

[0095] Example 2

[0096] See also Figure 2 As shown, based on the search method based on the software system provided in Example 1 of the present application, Example 2 of the present application proposes a search system based on the software system. Example 2 is only a preferred method of Example 1, and the implementation of Example 2 will not affect the independent implementation of Example 1.

[0097] Specifically, the software system-based search system provided in Embodiment 2 of the present application is different in that it includes a construction module, a processing module, a matching module, and an output module;

[0098] The building module is used to build an index structure for the collected data;

[0099] The processing module is used to convert the query request input by the current user into a preset query expression and execute the search query of the current user in the constructed index;

[0100] The matching module is used to locate each group of data that matches the current user query expression, and execute corresponding steps to obtain a filtered document set, and parse and evaluate each group of documents in the filtered document set through a pre-established relevance scoring model, so as to obtain a relevance evaluation index Gr of each group of documents in the current user filtered document set, where r represents the number of each group of documents in the filtered document set, r=1,2,...,k, k represents the total number of documents in the filtered document set, and sorts the documents based on the relevance evaluation index Gr of each group of documents in the filtered document set;

[0101] The output module is used to display the sorted search results in the format preset by the current user and push them to the current user. After receiving the pushed content, the current user can selectively perform corresponding steps to select the final document for viewing;

[0102] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A search method based on a software system, characterized in that: include: Building an index: preprocess the collected data and build an index structure based on the preprocessed data; Processing queries: converting the query request entered by the current user into a preset query expression; And execute the current user's search query in the built index; Search matching: Use the constructed index to locate each group of data that matches the current user's query expression, and execute the corresponding steps to obtain the filtered document set, parse and evaluate each group of documents in the filtered document set, so as to obtain the relevant evaluation index Gr of each group of documents in the current user's filtered document set, where r represents the number of each group of documents in the filtered document set, r = 1, 2, ..., k, k represents the total number of documents in the filtered document set, and sort based on the size of the relevant evaluation index Gr of each group of documents in the filtered document set; Output result: After sorting based on the size of the relevant evaluation index Gr of each group of documents in the filtered document collection, the sorted search results are displayed in the format preset by the current user and pushed to the current user. After receiving the pushed content, the current user can selectively perform the corresponding steps to select the final document for viewing.

2. A software system-based search method according to claim 1, characterized in that: Execute the corresponding steps to obtain the filtered document collection, specifically: S1: Extract each keyword in the current user query expression according to the constructed index structure, locate the document containing each keyword based on each extracted keyword, integrate the located groups of documents, and thus construct a document set corresponding to the current user query expression; S2: After building the document collection, parse the current user query expression and input the pre-built logical operator set for matching; the logical operators include "and", "or" and "not"; S3: If the match is successful, the logical operator of the successful match is parsed. If the parsing result is "and", the documents that meet each keyword at the same time are filtered out from the document collection to build a filtered document collection. If the parsing result is "or", the constructed document collection is directly used as the filtered document collection. If the parsing result is "no", the keyword corresponding to "no" is first identified. After the identification is completed, the documents that do not contain negative keywords are filtered out from the document collection to build a filtered document collection.

3. A software system-based search method according to claim 1, characterized in that: Each set of documents in the filtered document collection is parsed and evaluated, specifically: M1: Parse and filter the content of each group of documents in the document collection, remove negative keywords from each keyword in the current user query expression, match it with the content of each group of documents, count the number of keywords appearing in the content of each group of documents, divide the content of each group of documents into corresponding positions according to the title, abstract and content, and classify the number of keywords appearing in the content of each group of documents according to the divided positions, so as to obtain the number of keywords appearing in different divided positions of the content of each group of documents; set a weight coefficient corresponding to each divided position; The number of keywords appearing in different division positions of each group of document content is multiplied by the corresponding weight coefficient, and then summed up to obtain the criticality index e of each group of documents in the filtered document set; M2: Mark the time point when the current user inputs the query request as the starting point, preset three groups of time intervals, integrate the starting point with the preset three groups of time intervals, and obtain three groups of time windows for the current user, which are divided into recent time windows, medium-term time windows, and early time windows; obtain the release time of each group of documents in the filtered document collection, and match them with the three groups of time windows, set each group of time windows to correspond to a reference score, and use the reference score corresponding to each group of documents as the timeliness index r of each group of documents.

4. A software system-based search method according to claim 3, characterized in that: Parse and evaluate each set of documents in the filtered document collection, including: M3: Count the number of citations of each set of document content, and use the preset conversion rules to convert the number of citations of each set of document content into value scores; Preset the range of values ​​for each group of citations corresponding to the number of citations, and set each group of citations range to correspond to a value score; match the number of citations of each group of document content with the range of values ​​for each group of citations, so as to obtain the value score corresponding to each group of document content; Obtain the number of user feedback evaluations for each group of document content, extract the score corresponding to each user feedback evaluation from the number of user feedback evaluations for each group of document content, and take the average to obtain the average user score for each group of document content; The weight coefficients corresponding to the preset value scores and user score averages are respectively multiplied by the corresponding weight coefficients, and then the sum is calculated to obtain the authority index p of each group of document content.

5. A software system-based search method according to claim 4, characterized in that: Parse and evaluate each set of documents in the filtered document collection, including: M4: Send a network speed test data packet to the current user. The current user downloads after receiving the network speed test data packet. Record the time from the start of downloading to obtain the download speed value of the current user at each time point within the set time period. Use the standard deviation formula to calculate the download speed value of the current user at each time point within the set time period to obtain the network speed fluctuation value. Set a reference threshold for the network speed fluctuation value. If the network speed fluctuation value of the current user is greater than the set reference threshold, take the median value from the download speed values ​​at each time point as the network speed evaluation value. Otherwise, take the mean of the download speed values ​​at each time point as the network speed evaluation value. Based on the current user's network speed evaluation value, the current user's network speed is graded into low speed, medium speed and high speed, and a maximum value of the allowed download file is set for low speed, medium speed and high speed respectively; Set low speed, medium speed and high speed to correspond to a network speed range respectively, match the current user's network speed evaluation value with the corresponding network speed range, so as to complete the classification of user network speed; The file size of each group of document contents is obtained, and the ratio is calculated with the maximum value of the allowable downloadable file corresponding to the current user network speed evaluation value, so as to obtain the downloadability index m of each group of documents.

6. A software system-based search method according to claim 5, characterized in that: Get the relevant evaluation index Gr of each group of documents in the current user's filtered document collection, specifically: M5: Extract the criticality index e, timeliness index r, authority index p and downloadability index m corresponding to each set of document content, according to the formula where e 期望 、r 期望 and p 期望 They respectively represent the minimum expected criticality index, minimum expected timeliness index and minimum expected authority index pre-set by the current user; among them, a1, a2, a3 and a4 respectively represent the impact weight factors corresponding to the criticality index e, timeliness index r, authority index p and downloadability index m.

7. A software system-based search method according to claim 1, characterized in that: Selectively perform the corresponding steps to select the final document, specifically: Selection bias: After receiving the pushed content, if the user is not satisfied with the currently pushed sorted search results, the adjustment signaling is triggered and a selection interface is pushed to the current user. The selection interface includes key bias screening, timeliness bias screening, authority bias screening and download bias screening. After the current user selects the corresponding bias screening through the selection interface, the values ​​of the influence weight factors a1, a2, a3 and a4 of the key index e, timeliness index r, authority index p and download index m corresponding to each group of document content are adjusted according to the preset of the corresponding bias screening; Reselect: If the user is still not satisfied with the content currently pushed through the push selection interface, then after removing the negative keywords from each keyword in the current user query expression, synonym matching and conversion are performed, and steps S1-S3 and steps M1-M5 are executed again.

8. A software system-based search system, applied to a software system-based search method according to any one of claims 1 to 7, characterized in that: include: Construction module: build index structure for collected data; Processing module: converts the query request entered by the current user into a preset query expression, and executes the current user's search query in the constructed index; Matching module: locates each group of data matching the current user query expression, and executes corresponding steps to obtain a filtered document set, and parses and evaluates each group of documents in the filtered document set through a pre-established relevance scoring model, thereby obtaining a relevance evaluation index Gr of each group of documents in the current user filtered document set, where r represents the number of each group of documents in the filtered document set, r = 1, 2, ..., k, k represents the total number of documents in the filtered document set, and sorts the documents based on the relevance evaluation index Gr of each group of documents in the filtered document set; Output module: Display the sorted search results in the format preset by the current user and push them to the current user. After receiving the pushed content, the current user can selectively perform corresponding steps to select the final document for viewing.

Images