Document mixed search method, device and equipment

Abstract

The invention provides a document mixed search method, device and equipment. The document mixed search method comprises the following steps: acquiring a search word input by a user; according to the search word, performing word segmentation search in a database in which vectorized documents are stored to obtain a first search result; performing vectorization processing on the search word to obtain a search word vector; according to the search word vector, performing vector search in the database in which vectorized documents are stored to obtain a second search result; and processing the first search result and the second search result to obtain a target search document. According to the method, the document searching efficiency, the matching degree and the diversity of application scenes can be improved.

Description

Technical Field

[0001] The present invention relates to the field of computer information processing technology, and in particular to a document hybrid search method, device and equipment. Background Art

[0002] In the era of big data, search engines are gradually growing. From directory-based search engines to keyword-based search engines, these traditional search engines only use single words entered by users as keywords for some direct and simple matching, and the query results are not highly relevant. However, the keywords entered by users are no longer simple single words as before, but are searched in the form of short texts. The problems of Chinese word segmentation and relevance ranking of search results pose challenges to existing search engines. Summary of the invention

[0003] The technical problem to be solved by the present invention is to provide a document hybrid search method, device and equipment, which can improve the efficiency, matching degree and diversity of application scenarios of document search.

[0004] In order to solve the above technical problems, the technical solution of the present invention is as follows:

[0005] A hybrid document search method, comprising:

[0006] Get the search term entered by the user;

[0007] According to the search term, a word segmentation search is performed in a database storing vectorized documents to obtain a first search result;

[0008] Performing vectorization processing on the search term to obtain a search term vector;

[0009] Performing a vector search in the database storing vectorized documents according to the search term vector to obtain a second search result;

[0010] The first search result and the second search result are processed to obtain a target search document.

[0011] Optionally, based on the search term, a word segmentation search is performed in a database storing vectorized documents to obtain a first search result, including:

[0012] Searching the paragraphs of the vectorized document according to the search terms to obtain paragraph search results;

[0013] Performing a title search on the paragraph search results to obtain a title search result;

[0014] A first search result is obtained according to the title search result and the paragraph search result.

[0015] Optionally, obtaining a first search result according to the title search result and the paragraph search result includes:

[0016] The paragraph content in the paragraph search results, the document title and the document number in the title search results are taken as the first search results.

[0017] Optionally, vectorization is performed on the search term to obtain a search term vector, including:

[0018] Performing text cleaning on the search term to obtain a first search term;

[0019] According to the set parameters, the first search term is vectorized to obtain a search term vector.

[0020] Optionally, performing a vector search in the database storing vectorized documents according to the search term vector to obtain a second search result includes:

[0021] Obtaining the similarity of the document to the search term based on the search term vector and the vectorized document;

[0022] According to the similarity, a vector search is performed in the database storing the vectorized documents to obtain a second search result.

[0023] Optionally, processing the first search result and the second search result to obtain a target search document includes:

[0024] Deduplication of identical documents in the first search result and the second search result to obtain a first intermediate processing result;

[0025] The documents in the first intermediate processing result are sorted to obtain target search documents.

[0026] Optionally, sorting the documents in the first intermediate processing result to obtain the target search document includes:

[0027] Perform word segmentation on the search terms to obtain word segmentation results;

[0028] Using the word segmentation result, obtaining the number of word segmentation times the word segmentation result appears in the document;

[0029] The documents in the first intermediate processing result are sorted according to the number of word segmentation times to obtain target search documents.

[0030] An embodiment of the present invention further provides a document hybrid search device, comprising:

[0031] An acquisition module, used to acquire the search term input by the user;

[0032] A processing module is used to perform a word segmentation search in a database storing vectorized documents based on the search term to obtain a first search result; perform vectorization processing on the search term to obtain a search term vector; perform a vector search in the database storing vectorized documents based on the search term vector to obtain a second search result; and process the first search result and the second search result to obtain a target search document.

[0033] A computing device is also provided in an embodiment of the present invention, comprising: one or more processors; a storage device for storing one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the document hybrid search method described in the present invention.

[0034] An embodiment of the present invention further provides a computing device readable storage medium, including: a program is stored in the computing device readable storage medium, and when the program is executed by a processor, the document hybrid search method described in the present invention is implemented.

[0035] The above technical solution of the present invention has at least the following technical effects:

[0036] The above-mentioned hybrid document search method of the present invention obtains the search term input by the user; performs a word segmentation search in a database storing vectorized documents according to the search term to obtain a first search result; performs vector processing on the search term to obtain a search term vector; performs a vector search in the database storing vectorized documents according to the search term vector to obtain a second search result; and processes the first search result and the second search result to obtain a target search document. It can improve the efficiency, matching degree and diversity of application scenarios of document search. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 is a schematic diagram of the document hybrid search method of the present invention;

[0038] Figure 2 It is a schematic diagram of the document hybrid search device of the present invention. DETAILED DESCRIPTION

[0039] The exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present invention and to enable the scope of the present invention to be fully communicated to those skilled in the art.

[0040] like Figure 1 As shown, an embodiment of the present invention proposes a document hybrid search method, comprising:

[0041] Step S1, obtaining the search term input by the user;

[0042] Step S2, performing a word segmentation search in a database storing vectorized documents according to the search term to obtain a first search result;

[0043] Step S3, performing vectorization processing on the search term to obtain a search term vector;

[0044] Step S4, performing a vector search in the database storing vectorized documents according to the search term vector to obtain a second search result;

[0045] Step S5: Process the first search result and the second search result to obtain a target search document.

[0046] In this embodiment, the vectorized documents stored in the database are determined and stored in the database through the following process:

[0047] Acquire the layout document content, segment the layout document content according to natural paragraphs to obtain multiple paragraphs, merge the data of multiple paragraphs, and the size of the merged data after each merge is less than or equal to a preset value, which can be 512 words, so as to ensure the efficiency of the search. If the merged data is too large, the search performance will be reduced; vectorize each merged data to obtain multiple vectorized data, and store the document title, paragraph content, document identifier, and vector data of the layout document in a database, which can be a non-relational database, such as an ES (Elasticsearch) database.

[0048] The vectorization processing of each merged data may include:

[0049] Get the real vector of each word in the merged data; map the real vector of each word to the real vector space. Each word is represented as a real vector of fixed length, which represents the semantic information of the word in a high-dimensional vector space. For example, the words: "cat" and "dog". They each have a vector representation, the vector of "cat": 0.2, -0.5, 0.8; the vector of "dog": 0.6, 0.1, -0.7;

[0050] The numbers in the vector represent the semantic features of the word. The word vector can represent the similarity between words. For example, "cat" and "dog" may be closer in semantics, and their vectors are closer in this vector space. This representation method helps to better understand and generate natural language, as well as better search for target documents related to words.

[0051] like Figure 1As shown, the user inputs the search term to be searched into the search engine, and the search engine first obtains the input search term, performs word segmentation processing on the search term, and uses the word segmentation processing result to perform word segmentation search in a database storing vectorized documents to obtain a first search result; next, the obtained search term is vectorized, and the vectorized search term vector is used to perform a vector search in the vectorized document database to obtain a second search result; finally, the first search result and the second search result are compared, and the repeated parts in the first search result and the second search result are removed, and the search results after deduplication are re-sorted to obtain the target search result; the solution of the present invention uses word segmentation search and vector search in a mixed way, which not only utilizes the advantage of the higher occurrence rate of word segmentation search search terms, but also utilizes the advantage of the higher semantic similarity of vector search search terms. The two methods are combined to improve the efficiency and quality of document retrieval, and can meet the diversity of user usage scenarios.

[0052] In an optional embodiment of the present invention, in step S2, a word segmentation search is performed in a database storing vectorized documents according to the search term to obtain a first search result, including:

[0053] Step S21, searching the paragraphs of the vectorized document according to the search term to obtain paragraph search results;

[0054] Step S22, performing a title search on the paragraph search results to obtain a title search result;

[0055] Step S23, obtaining a first search result according to the title search result and the paragraph search result.

[0056] In this embodiment, the search words are constructed according to the type of the query string of the database (for example, keyword type or text type, keywords are stored as a whole when stored, and text type data will be segmented and stored when storing data) to construct a query fuzzy query statement;

[0057] If the search term is a keyword, a fuzzy query is performed according to the keyword's fuzzy query statement (after segmenting the search term, query according to the segmentation result). For example, if the search term is "The weather is great today", the segmented results are "today", "weather", and "great". However, keyword-type field data will not be segmented, so it also needs to be fully matched to be queried. The fuzzy query will search for data in the database's Name field that only contains "today", "weather", and "great", and then return the matched data in reverse order of matching degree.

[0058] If the search term is of text type, a fuzzy query is performed according to the fuzzy query statement of the text type. For example, if the search term is: read a, the fuzzy query first segmentes the search term to obtain read and a, and then matches the data containing these two strings after segmentation. For example, the returned data is read a book and you can get a good job, because these two data contain the string a. Since read a book satisfies both search conditions (read and a), read a book will have a higher matching degree and is returned as the first data.

[0059] When searching by search terms, you can also match phrases. For example, for data with address="read a book", searching for read a, a book, or read a book will filter out this data.

[0060] According to the search term, the paragraphs of the vectorized document are searched, and when the paragraph search results are obtained, the search term is firstly segmented using a word segmentation tool, and then according to the above-mentioned type of word segmentation, a fuzzy search is performed on the paragraphs of the vectorized document to obtain the paragraph search results;

[0061] Secondly, based on the word segmentation, a title search is performed on the paragraph search results to obtain a title search result;

[0062] Finally, set the matching word weights, and weight the matching words that appear in the title. Preferably, set the weight multiple to 50; next set the minimum matching degree of the search, and results with lower matching degrees do not need to be retrieved, for example: set the minimum matching degree to less than 90% when the matching word appears 3 times, less than 70% when the matching word appears 5 times, and less than 30% when the matching word appears 7 times. The search term is "The weather is so nice today", and the search term is segmented into "today", "weather", and "really nice". If there are less than 2 matching words during the search, the document will not be displayed in the search list; weight and minimum matching degree processing are performed on the title search results to obtain the first search result.

[0063] In an optional embodiment of the present invention, in step S23, obtaining a first search result according to the title search result and the paragraph search result includes:

[0064] The paragraph content in the paragraph search results, the document title and the document number in the title search results are taken as the first search results.

[0065] In this embodiment, in order to better present the search results, the content of the first search result includes the paragraph content in the paragraph search results, the document title and the document number in the title search results.

[0066] In an optional embodiment of the present invention, in step S3, vectorization is performed on the search term to obtain a search term vector, including:

[0067] Step S31, performing text cleaning on the search term to obtain a first search term;

[0068] Step S32: vectorize the first search term according to the set parameters to obtain a search term vector.

[0069] In this embodiment, the search term is vectorized. First, the search term is cleaned to remove irrelevant characters, punctuation marks, special symbols, etc. in the text, and useful search term information is retained to obtain a first search term. Then, parameters such as vector dimension, context window size, and number of iterations are determined to vectorize the first search term. The vectorization process here is the same as the vectorization process of the vector document in the database. The first search term is vectorized with real numbers to obtain a search term vector.

[0070] In an optional embodiment of the present invention, in step S4, a vector search is performed in the database storing vectorized documents according to the search term vector to obtain a second search result, including:

[0071] Step S41, obtaining the similarity of the document to the search term based on the search term vector and the vectorized document;

[0072] Step S42: performing a vector search in the database storing the vectorized documents according to the similarity to obtain a second search result.

[0073] In this embodiment, the document closest to the query word vector is searched in the space of vectorized documents in the database to obtain a second search result with semantic similarity to the query word vector; specifically, the vectorized document is searched using the K nearest neighbor algorithm, the entire training data set is traversed first, the distance between the query point and each point in the training data set is calculated, and the K points closest to the query point are selected as the nearest neighbors according to the calculated distance. After the nearest neighbors are found, a majority vote is performed according to the categories of these neighbors to determine the matching degree of the query point; specifically, the cosine similarity metric is used to calculate the similarity between vectors, for example, to calculate the similarity between two vectors A (x1, x2, ... x n ) and B(y1, y2,…y n ) Cosine similarity, calculated as follows:

[0074]

[0075] Among them, Sim (A, B) represents the cosine similarity of two vectors A and B. The larger the cosine similarity, the higher the similarity of the two vectors. Cosine similarity measures similarity from the direction and is more suitable for text similarity measurement.

[0076] Then, according to the similarity, the search results with higher similarity are output to obtain the second search results.

[0077] In an optional embodiment of the present invention, in step S5, the first search result and the second search result are processed to obtain a target search document, including:

[0078] Step S51, removing duplicate documents in the first search result and the second search result to obtain a first intermediate processing result;

[0079] Step S52: sort the documents in the first intermediate processing result to obtain target search documents.

[0080] In this embodiment, the search results are deduplicated and re-sorted. First, the document labels in the first search result and the second search result are compared. For documents with the same document labels, the document with the identification number in the second search result is deleted, and the document with the identification number in the first search result is retained to obtain a first intermediate processing result. Then, the documents in the first intermediate processing result are sorted to obtain the target search document.

[0081] In an optional embodiment of the present invention, in step S52, sorting the documents in the first intermediate processing result to obtain the target search document includes:

[0082] Step S521, performing word segmentation processing on the search term to obtain a word segmentation result;

[0083] Step S522, using the word segmentation result, obtaining the number of times the word segmentation result appears in the document;

[0084] Step S523, sorting the documents in the first intermediate processing result according to the number of word segmentation times to obtain the target search document.

[0085] In this embodiment, the documents in the first intermediate processing result are sorted. First, the search terms are segmented using a word segmentation tool. The number of times the word segmentation appears in the document title and paragraph text is counted in the first intermediate processing result, and the weight of the number of times the word segmentation appears in the title is set. The number of document highlights is calculated using the formula Z=X*N+Y, where Z is the number of document highlights, X is the number of times the word segmentation appears in the title, N is the title weight, and Y is the number of times the analysis appears in the paragraph text. Finally, the first intermediate processing result is output and highlighted in the order of the number of document highlights from high to low to obtain the target search results.

[0086] The specific implementation process of the above method of the present invention is described below:

[0087] Step 111: Obtain the search term input by the user in the search interface;

[0088] Step 112: Search the paragraphs of the vectorized document according to the search term to obtain paragraph search results;

[0089] Step 113, performing a title search on the paragraph search results to obtain a title search result;

[0090] Step 114: Obtain a first search result based on the title search result and the paragraph search result;

[0091] Step 115: Use the paragraph content in the paragraph search results, the document title and the document number in the title search results as the first search results;

[0092] Step 116: Perform text cleaning on the search term to obtain a first search term;

[0093] Step 117: vectorize the first search term according to the set parameters to obtain a search term vector;

[0094] Step 118: Obtain the similarity of the document to the search term based on the search term vector and the vectorized document vector;

[0095] Step 119: Obtain a second search result according to the similarity;

[0096] Step 120: remove duplicate documents in the first search result and the second search result to obtain a first intermediate processing result;

[0097] Step 121, performing word segmentation processing on the search term to obtain a word segmentation result;

[0098] Step 122: using the word segmentation result, obtain the number of times the word segmentation result appears in the document;

[0099] Step 124: Sort the documents in the first intermediate processing result according to the number of word segmentations to obtain target search documents.

[0100] The above-mentioned scheme of the present invention uses word segmentation search and vector search in a mixed way, which not only utilizes the advantage of the higher occurrence rate of search terms in word segmentation search, but also utilizes the advantage of the higher semantic similarity of search terms in vector search. The two methods are combined with each other to improve the efficiency and quality of document retrieval and can meet the diversity of user usage scenarios.

[0101] like Figure 2As shown, an embodiment of the present invention further provides a document hybrid search device 20, comprising:

[0102] An acquisition module 21 is used to acquire a search term input by a user;

[0103] The processing module 22 is used to perform a word segmentation search in a database storing vectorized documents based on the search term to obtain a first search result; perform vectorization processing on the search term to obtain a search term vector; perform a vector search in the database storing vectorized documents based on the search term vector to obtain a second search result; and process the first search result and the second search result to obtain a target search document.

[0104] Optionally, based on the search term, a word segmentation search is performed in a database storing vectorized documents to obtain a first search result, including:

[0105] Searching the paragraphs of the vectorized document according to the search terms to obtain paragraph search results;

[0106] Performing a title search on the paragraph search results to obtain a title search result;

[0107] A first search result is obtained according to the title search result and the paragraph search result.

[0108] Optionally, obtaining a first search result according to the title search result and the paragraph search result includes:

[0109] The paragraph content in the paragraph search results, the document title and the document number in the title search results are taken as the first search results.

[0110] Optionally, vectorization is performed on the search term to obtain a search term vector, including:

[0111] Performing text cleaning on the search term to obtain a first search term;

[0112] According to the set parameters, the first search term is vectorized to obtain a search term vector.

[0113] Optionally, performing a vector search in the database storing vectorized documents according to the search term vector to obtain a second search result includes:

[0114] Obtaining the similarity of the document to the search term based on the search term vector and the vectorized document;

[0115] According to the similarity, a vector search is performed in the database storing the vectorized documents to obtain a second search result.

[0116] Optionally, processing the first search result and the second search result to obtain a target search document includes:

[0117] Deduplication of identical documents in the first search result and the second search result to obtain a first intermediate processing result;

[0118] The documents in the first intermediate processing result are sorted to obtain target search documents.

[0119] Optionally, sorting the documents in the first intermediate processing result to obtain the target search document includes:

[0120] Perform word segmentation on the search terms to obtain word segmentation results;

[0121] Using the word segmentation result, obtaining the number of word segmentation times the word segmentation result appears in the document;

[0122] The documents in the first intermediate processing result are sorted according to the number of word segmentation times to obtain target search documents.

[0123] It should be noted that all implementation methods in the above method embodiments are applicable to the embodiments of the device and can achieve the same technical effects.

[0124] An embodiment of the present invention further provides a computing device, comprising:

[0125] one or more processors;

[0126] A storage device is used 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 document hybrid search method of the present invention. All implementation methods in the above method embodiments are applicable to the embodiments of the device and can achieve the same technical effect.

[0127] The embodiment of the present invention further provides a computing device readable storage medium, wherein the computing device readable storage medium stores a program, and when the program is executed by a processor, the document hybrid search method of the present invention is implemented. All implementation methods in the above method embodiment are applicable to the embodiment of the computing device readable storage medium, and can also achieve the same technical effect.

[0128] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0129] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0130] In the embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0131] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0132] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0133] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, ROM, RAM, magnetic disks, or optical disks.

[0134] In addition, it should be pointed out that in the apparatus and method of the present invention, it is obvious that each component or each step can be decomposed and / or recombined. These decompositions and / or recombinations should be regarded as equivalent schemes of the present invention. Moreover, the steps of performing the above series of processing can naturally be performed in chronological order according to the order of description, but it is not necessary to perform them in chronological order, and some steps can be performed in parallel or independently of each other. For those of ordinary skill in the art, it is understandable that all or any steps or components of the method and apparatus of the present invention can be implemented in hardware, firmware, software or a combination thereof in any computing device (including processors, storage media, etc.) or a network of computing devices, which can be achieved by those of ordinary skill in the art using their basic programming skills after reading the description of the present invention.

[0135] Therefore, the purpose of the present invention can also be achieved by running a program or a group of programs on any computing device. The computing device can be a well-known general-purpose device. Therefore, the purpose of the present invention can also be achieved by simply providing a program product containing a program code for implementing a method or device. That is to say, such a program product also constitutes the present invention, and a storage medium storing such a program product also constitutes the present invention. Obviously, the storage medium can be any well-known storage medium or any storage medium developed in the future. It should also be pointed out that in the device and method of the present invention, it is obvious that each component or each step can be decomposed and / or recombined. These decompositions and / or recombinations should be regarded as equivalent schemes of the present invention. In addition, the steps of performing the above-mentioned series of processing can naturally be performed in chronological order according to the order of description, but it is not necessary to perform them in chronological order. Some steps can be performed in parallel or independently of each other.

[0136] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A document hybrid search method, characterized in that: include: Get the search term entered by the user; According to the search term, a word segmentation search is performed in a database storing vectorized documents to obtain a first search result; Performing vectorization processing on the search term to obtain a search term vector; Performing a vector search in the database storing vectorized documents according to the search term vector to obtain a second search result; The first search result and the second search result are processed to obtain a target search document.

2. The document hybrid search method according to claim 1, characterized in that: According to the search term, a word segmentation search is performed in a database storing vectorized documents to obtain a first search result, including: Searching the paragraphs of the vectorized document according to the search terms to obtain paragraph search results; Performing a title search on the paragraph search results to obtain a title search result; A first search result is obtained according to the title search result and the paragraph search result.

3. The document hybrid search method according to claim 2, characterized in that: According to the title search result and the paragraph search result, a first search result is obtained, including: The paragraph content in the paragraph search results, the document title and the document number in the title search results are taken as the first search results.

4. The document hybrid search method according to claim 1, characterized in that: The search term is vectorized to obtain a search term vector, including: Performing text cleaning on the search term to obtain a first search term; According to the set parameters, the first search term is vectorized to obtain a search term vector.

5. The document hybrid search method according to claim 1, characterized in that: According to the search term vector, a vector search is performed in the database storing the vectorized documents to obtain a second search result, including: Obtaining the similarity of the document to the search term based on the search term vector and the vectorized document; According to the similarity, a vector search is performed in the database storing the vectorized documents to obtain a second search result.

6. The document hybrid search method according to claim 1, characterized in that: Processing the first search result and the second search result to obtain a target search document includes: Deduplication of identical documents in the first search result and the second search result to obtain a first intermediate processing result; The documents in the first intermediate processing result are sorted to obtain target search documents.

7. The document hybrid search method according to claim 6, characterized in that: Sorting the documents in the first intermediate processing result to obtain target search documents includes: Perform word segmentation on the search terms to obtain word segmentation results; Using the word segmentation result, obtaining the number of word segmentation times the word segmentation result appears in the document; The documents in the first intermediate processing result are sorted according to the number of word segmentation times to obtain target search documents.

8. A hybrid document search device, characterized in that: include: An acquisition module, used to acquire the search term input by the user; A processing module is used to perform a word segmentation search in a database storing vectorized documents based on the search term to obtain a first search result; perform vectorization processing on the search term to obtain a search term vector; perform a vector search in the database storing vectorized documents based on the search term vector to obtain a second search result; and process the first search result and the second search result to obtain a target search document.

9. A computing device, characterized in that include: one or more processors; A storage device for storing 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 method as claimed in any one of claims 1 to 7.

10. A computing device readable storage medium, characterized in that: The computing device readable storage medium stores a program, which implements the method according to any one of claims 1 to 7 when executed by a processor.

Images