A method and apparatus for complementing search content
By extracting key information from the titles of documents to be recalled, establishing a set of suggested terms, and calculating recall scores, the problem of low recall rate in traditional search autocomplete algorithms is solved, resulting in more efficient search result display and improved user experience.
Patent Information
- Application Number
- CN202110676021.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-06-18
- Publication Date
- 2026-01-20
- Estimated Expiration
- 2041-06-18
AI Technical Summary
Traditional search autocomplete algorithms rely on prefix matching based on the user's search terms, which fails to effectively recall matching results in documents, resulting in low recall rates.
By extracting key information from the titles of documents to be recalled, a set of suggested words is established, and a prefix matching algorithm and a text vector normalization algorithm are used to determine the recall score, thereby optimizing the ranking and display of search results.
It improved the recall rate of search autocomplete, provided a variety of suggested words, and enhanced the user's search experience.
Smart Images
Figure CN115495548B_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of intelligent search, in particular to a method and device for completing search content. BACKGROUND
[0002] A traditional search auto-completion algorithm is based on a user's search word, performs prefix matching in a dictionary tree (Trie tree) of a set of to-be-recalled documents, returns a matching result satisfying a prefix word, and constructs a prompt word sequence of a search auto-completion function.
[0003] The prefix matching algorithm based on the Trie tree can maximize the use of information contained in the user's search word, improve the accuracy and search speed, but cannot perform matching from the middle of the to-be-recalled document. Since the prefix matching algorithm starts matching from the first word, when the first word of the user's search word is different from the first word of the to-be-recalled document, the document will be rejected to be recalled. For example, the user inputs "Ma cake", and the to-be-recalled document contains "Macaron dessert", "French Macaron dessert", "Italian Macaron dessert", and "French Macaron". According to the traditional search auto-completion algorithm, only "Macaron dessert" can be recalled and displayed in the prompt word sequence of the search auto-completion. Therefore, the recall rate of the traditional search auto-completion algorithm needs to be improved. SUMMARY
[0004] The present application provides a method and device for completing search content, which are used for matching a user's input search word with key information in a title of a to-be-recalled document, determining a corresponding title of the to-be-recalled document according to a matching degree, and improving the recall rate of the auto-completion algorithm.
[0005] In a first aspect, the present application provides a method for completing search content, which includes the following steps.
[0006] From a set of suggestion words corresponding to each to-be-recalled document, a suggestion word matched with the user's input search word is determined, wherein the set of suggestion words is used to represent key information in a title of the corresponding to-be-recalled document.
[0007] According to a matching degree of the matched suggestion word and the search word, a recall score of the suggestion word is determined.
[0008] According to the recall score of the suggestion word, a title of a to-be-recalled document corresponding to a preset number of highest-score suggestion words is determined, and the title is displayed.
[0009] The completion method provided by the embodiment of the present application can extract key information of a document title to be recalled, compare the search word input by the user with the key information of the title, determine a matched title, recall the matched document, and display the title of the document. Since the key information in the title to be recalled is extracted, the key information in the title is used to match the search word, and the recall rate of the document is improved.
[0010] In some examples, the set of suggestion words corresponding to the document to be recalled is determined in the following manner:
[0011] The title of the document to be recalled is segmented to obtain various title fields;
[0012] Key words of the various title fields are extracted, and the set of suggestion words corresponding to the document to be recalled is determined according to all the extracted key words.
[0013] In some examples, the suggestion word matched with the search word input by the user is determined from the set of suggestion words corresponding to each document to be recalled in the following manner:
[0014] A dictionary tree of the document to be recalled is established based on the various suggestion words in the set of suggestion words corresponding to the document to be recalled;
[0015] The suggestion words in the dictionary tree are matched by a prefix matching algorithm to determine the suggestion word matched with the search word.
[0016] In some examples, the recall score of the suggestion word is determined according to the matching degree of the matched suggestion word and the search word in the following manner:
[0017] The weight of the suggestion word is determined according to the matching degree of the matched suggestion word and the search word;
[0018] The weight of the suggestion word is adjusted based on a text vector normalization algorithm to obtain the recall score of the suggestion word.
[0019] In some examples, the title is displayed in the following manner:
[0020] The document to be recalled corresponding to the suggestion word is sorted according to the recall score of the suggestion word;
[0021] The title of the document to be recalled is displayed in the sorted order.
[0022] In a second aspect, the embodiment of the present application provides a device for completing search content, which comprises a processor and a memory. The memory is used to store programs executable by the processor, and the processor is used to read the programs in the memory and perform the following steps:
[0023] determine a suggestion word matching the search word from each suggestion word set corresponding to a document to be recalled, wherein the suggestion word set is used to represent key information in a title of the corresponding document to be recalled;
[0024] determine a recall score of the suggestion word according to a matching degree of the matching suggestion word and the search word;
[0025] determine a title of a document to be recalled corresponding to a suggestion word with a highest score from the preset number of suggestion words according to the recall score of the suggestion word, and display the title.
[0026] In some examples, the processor is specifically configured to determine the suggestion word set corresponding to the document to be recalled by:
[0027] segment the title of the document to be recalled to obtain each title field;
[0028] extract key words of the each title field, and determine the suggestion word set corresponding to the document to be recalled according to all the extracted key words.
[0029] In some examples, the processor is specifically configured to perform:
[0030] establish a dictionary tree of the document to be recalled based on each suggestion word in the suggestion word set corresponding to the document to be recalled;
[0031] determine the suggestion word matching the search word by prefix matching algorithm on the suggestion words in the dictionary tree.
[0032] In some examples, the processor is specifically configured to perform:
[0033] determine a weight of the suggestion word according to a matching degree of the matching suggestion word and the search word;
[0034] adjust the weight of the suggestion word based on a text vector normalization algorithm to obtain the recall score of the suggestion word.
[0035] In some examples, the processor is specifically configured to perform:
[0036] sort the document to be recalled corresponding to the suggestion word according to the recall score of the suggestion word;
[0037] display the title of the document to be recalled in the order of the sorting.
[0038] In a third aspect, an apparatus for completing search content is provided, and the apparatus comprises:
[0039] determining a suggestion word unit configured to determine a suggestion word matched with the search word from a suggestion word set corresponding to each document to be recalled, wherein the suggestion word set is used to represent key information in a title of the corresponding document to be recalled;
[0040] determining a recall score unit configured to determine a recall score of the suggestion word according to a matching degree of the matched suggestion word and the search word;
[0041] determining a title display unit configured to determine a title of a document to be recalled corresponding to a suggestion word with a highest score according to the recall scores of the suggestion words, and display the title.
[0042] In some examples, the determining suggestion word unit is specifically configured to determine the suggestion word set corresponding to the document to be recalled by:
[0043] segmenting the title of the document to be recalled to obtain each title field;
[0044] extracting key words of the each title field, and determining the suggestion word set corresponding to the document to be recalled according to all the extracted key words.
[0045] In some examples, the determining suggestion word unit is specifically configured to:
[0046] establishing a dictionary tree of the document to be recalled based on each suggestion word in the suggestion word set corresponding to the document to be recalled;
[0047] determining the suggestion word matched with the search word by prefix matching algorithm on the suggestion words in the dictionary tree.
[0048] In some examples, the determining recall score unit is specifically configured to:
[0049] determining a weight of the suggestion word according to the matching degree of the matched suggestion word and the search word;
[0050] adjusting the weight of the suggestion word based on a text vector normalization algorithm to obtain the recall score of the suggestion word.
[0051] In some examples, the determining title display unit is specifically configured to:
[0052] sorting the document to be recalled corresponding to the suggestion word according to the recall score of the suggestion word;
[0053] displaying the title of the document to be recalled in the order of the sorting.
[0054] Fourthly, embodiments of the present invention also provide a computer storage medium having a computer program stored thereon, which, when executed by a processor, is used to implement the steps of the method described in the first aspect above.
[0055] These or other aspects of this application will become more apparent in the following description of embodiments. Attached Figure Description
[0056] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0057] Figure 1 A flowchart illustrating an implementation of content completion search provided in this embodiment of the invention;
[0058] Figure 2 This is a flowchart illustrating a method for preprocessing documents to be recalled, as provided in an embodiment of the present invention.
[0059] Figure 3 This is a flowchart illustrating an implementation method for automatically completing search terms according to an embodiment of the present invention.
[0060] Figure 4 This is a schematic diagram illustrating the display of completed search content according to an embodiment of the present invention.
[0061] Figure 5 A device for completing search content is provided in an embodiment of the present invention;
[0062] Figure 6 This invention provides an apparatus for completing search content. Detailed Implementation
[0063] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.
[0064] In this embodiment of the invention, the term "and / or" describes the relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. The character " / " generally indicates that the preceding and following associated objects have an "or" relationship.
[0065] The application scenarios described in the embodiments of this invention are for the purpose of more clearly illustrating the technical solutions of the embodiments of this invention, and do not constitute a limitation on the technical solutions provided by the embodiments of this invention. Those skilled in the art will understand that with the emergence of new application scenarios, the technical solutions provided by the embodiments of this invention are also applicable to similar technical problems. In the description of this invention, unless otherwise stated, "multiple" means two or more.
[0066] Traditional search autocomplete algorithms are based on user search terms and utilize search engines to select potentially relevant documents from a large dataset—a process called recall. Specifically, this involves prefix matching within a Trie tree (a tree of documents to be recalled), returning matches that satisfy the prefix, and constructing a sequence of suggested terms for autocomplete. The documents to be recalled are all documents under a specific index of the search engine. The Trie is built based on a field (such as a prefix field) of the documents, providing high-performance prefix matching for autocomplete. While Trie-based prefix matching algorithms maximize the use of information contained in user search terms, improving accuracy and search speed, they cannot match from the middle of the documents to be recalled. Because prefix matching starts from the first character, documents whose first characters do not match the user's search terms will be rejected. To address this technical problem, this embodiment utilizes automatic keyword recognition technology to identify key information in document titles, adding suggested words for search purposes. A prefix matching algorithm is then used to match suggested words in the suggested word set. This improves document recall during auto-completion while satisfying user-input search terms, providing diverse suggested words, optimizing the sorting order of suggested words, and enhancing the user's search experience.
[0067] The core idea of this invention is to segment the title of the document to be recalled to obtain key information of the document title, add suggested words for matching with search terms to the document to be recalled, and match not only the document title, but also the key information of the title. The recall score is determined by the degree of matching, and the titles of the documents to be recalled corresponding to the suggested words with the highest scores are selected according to the scores, thereby effectively improving the recall rate and enhancing the user search experience.
[0068] In some examples, such as Figure 1 As shown in the figure, the implementation process of a method for completing search content provided by an embodiment of the present invention is as follows:
[0069] Step 100: From the set of suggested words corresponding to each document to be recalled, determine the suggested words that match the search terms entered by the user, wherein the set of suggested words is used to represent the key information in the title of the corresponding document to be recalled;
[0070] In this embodiment, the suggested word set corresponding to each document to be recalled contains key information from the title of the document. In practice, the title of the document to be recalled can be segmented to extract key information. For example, if the title is "Macaron Desserts," the suggested word set corresponding to this title includes "Macaron Desserts" and "Desserts." In this embodiment, when matching search terms, the suggested word set generated from the key information of the title is not matched against the title of the document to be recalled. Since the suggested word set contains all the key information in the title, i.e., the key fields, matching using the suggested word set can improve the recall rate.
[0071] Step 101: Determine the recall score of the suggested term based on the degree of matching between the matched suggested term and the search term;
[0072] Step 102: Based on the recall scores of the suggested words, determine the titles of the documents to be recalled corresponding to a preset number of suggested words with the highest scores, and display the titles.
[0073] This embodiment can also optimize the matched suggested words, filter them according to the recall score, and display the titles of the documents to be recalled corresponding to the preset number of highest-scoring suggested words.
[0074] In some examples, this embodiment determines the set of suggested words corresponding to the document to be recalled in the following way:
[0075] The title of the document to be recalled is segmented to obtain individual title fields;
[0076] Extract keywords from each title field, and determine the set of suggested words corresponding to the document to be recalled based on all extracted keywords.
[0077] During implementation, the keywords for each title field are shown in Table 1:
[0078] Keywords in the title of Table 1
[0079] Title Suggested word set Macaron dessert Macaron dessert, dessert French macaron dessert French macaron dessert, macaron dessert, dessert Italian macaron dessert Italian macaron dessert, macaron dessert, dessert French macaron French macaron, macaron
[0080] In some examples, this embodiment also establishes information about documents to be recalled. When performing search term matching, based on the correspondence between the documents to be recalled, document content, document title, and suggested term set, the matching documents to be recalled are retrieved, and the titles of the recalled documents are displayed to automatically complete the search terms entered by the user. The content of the document information is shown in Table 2.
[0081] Table 2 Document Information
[0082] ID Title Content 10001 Macaron dessert Macaron dessert is from France... 10002 French macaron dessert In France, there is a famous dessert: macaron,... 10003 Italian macaron dessert Italy has macarons too! Not only... 10004 French macaron Macaron is a dessert that goes with afternoon tea...
[0083] In implementation, the information of the document to be recalled in this embodiment includes, but is not limited to: document identifier, document content, document title, and a set of suggested words for the document, and there is a one-to-one correspondence between the document identifier, document content, document title, and set of suggested words for the document. The correspondence is shown in Table 3.
[0084] Table 3. Correspondence between document information and weights
[0085]
[0086] In some examples, this embodiment also provides a suggestion word update method. When it is determined that the document to be recalled needs to be updated, the method checks whether the key information in the title of all the documents to be recalled already exists in the suggestion word set corresponding to the document to be recalled. If it is determined that new key information exists, the key information is added to the suggestion word set.
[0087] In some examples, embodiments of the present invention determine suggested terms that match the user-input search terms from the suggested term set corresponding to each document to be recalled in the following manner:
[0088] Based on each suggested term in the suggested term set corresponding to the document to be recalled, a trie (prefix tree) is built for the document to be recalled. A prefix matching algorithm is then used to perform prefix matching on the suggested terms in the trie to determine the suggested terms that match the search term. Optionally, a search engine can be used to build the trie and perform prefix matching on the document to be recalled. Alternatively, the information of the document to be recalled, containing the suggested term set, can be imported into the relevant index in Elasticsearch, thereby leveraging the ES search engine to achieve near real-time global retrieval and improve search response speed.
[0089] In some examples, this embodiment extracts key information from the title to generate a set of suggested words, and sets a weight for each suggested word for subsequent calculation of the recall score of the matched suggested words:
[0090] Method 1: Based on rules of word matching, such as TF-IDF or BM25 (Best Matching 25), segment the title and set the weight of suggested words;
[0091] Method 2: A machine learning-based approach is used to identify key information in the titles of documents to be recalled and to assign weights to each key piece of information.
[0092] In some examples, this embodiment determines the recall score of the suggested term based on the degree of matching between the matched suggested term and the search term in the following manner:
[0093] 1) Determine the weight of the suggested term based on the degree of matching between the suggested term and the search term;
[0094] The weights can be set based on text length normalization, i.e., weight = field length of suggested words / field length of title. For example, if the suggested word set for the title "Macaron Desserts" includes "Macaron Desserts" and "Desserts", then the length of "Macaron Desserts" in the suggested word set is the same as the original title length, and the weight of "Macaron Desserts" in this suggested word set is 1. Similarly, the weight of "Desserts" in this suggested word set is 2 / 5.
[0095] In practice, the recommended keyword weight settings are shown in Table 4.
[0096] Table 4 Suggested Keyword Weighting Settings
[0097] Title Suggested word set Weight setting Macaron dessert Macaron dessert, dessert (1，2 / 5) French macaron dessert French macaron dessert, macaron dessert, dessert (1，5 / 7，2 / 7) Italian macaron dessert Italian macaron dessert, macaron dessert, dessert (1，5 / 7，3 / 7) French macaron French macaron, macaron (1，3 / 5) French baguette French baguette, baguette (1，3 / 5) Macaron cake Macaron cake, cake (1，2 / 5)
[0098] 2) The weights of the suggested words are adjusted based on the text vector normalization algorithm to obtain the recall score of the suggested words.
[0099] In implementation, a text vector normalization algorithm can be used to calculate a normalized score for each suggested term corresponding to the title. The product of the suggested term's weight and the normalized score is then used to determine the recall score for that suggested term. The calculation rule for the normalized score of suggested terms obtained based on the text vector normalization algorithm is as follows: field length of the user-input search term / field length of the matched suggested term.
[0100] In practice, if the user enters the search term "maca", the normalized scores of the matching suggestions are shown in Table 5:
[0101] Table 5 Normalized scores of suggested terms
[0102] Title Matched suggested word Normalized score Macaron dessert Macaron dessert (2 / 5) Macaron cake Macaron cake (2 / 5) French macaron dessert Macaron dessert (2 / 5) Italian macaron dessert Macaron dessert (2 / 5) French macaron Macaron (2 / 3)
[0103] Based on the weights of each suggested term in Table 4 and the normalized scores of the matched suggested terms in Table 5, if the user's search term is "Maca", the weights and normalized scores of the suggested terms are multiplied together to obtain the recall scores of the matched suggested terms, as shown in Table 6.
[0104] Table 6 shows the weights and normalized scores of the suggested terms.
[0105] Title Weight Matched suggested word Normalized score Macaron dessert 1 Macaron dessert 1*(2 / 5)＝0.4 Macaron cake 1 Macaron cake 1*(2 / 5)0.4 French macaron dessert 5 / 7 Macaron dessert (5 / 7)*(2 / 5)＝0.286 Italian macaron dessert 5 / 7 Macaron dessert (5 / 7)*(2 / 5)＝0.286 French macaron 3 / 5 Macaron (3 / 5)*(2 / 3)＝0.4
[0106] In some examples, this embodiment displays the titles as follows: Based on the recall score of the suggested keywords, the documents to be recalled corresponding to the suggested keywords are sorted; the titles of the documents to be recalled are displayed in the sorted order. In practice, the titles of the corresponding documents to be recalled can be displayed sequentially in descending order of recall score.
[0107] In some examples, such as Figure 2 As shown in the figure, this embodiment of the invention provides a method for preprocessing documents to be recalled. The specific implementation flow of this method is as follows:
[0108] Step 200: Construct the correspondence between the document identifier, document content, and document title of the document to be recalled;
[0109] Step 201: Extract key information for each document to be recalled, generate a set of suggested words using the key information, and set the weight of each suggested word;
[0110] Step 202: Construct the correspondence between the document identifier, document content, document title, and suggested word set of the document to be recalled, and generate the document information to be recalled based on the correspondence;
[0111] Step 203: Import the document information to be recalled into the search engine's index for use in suggesting terms to match user-input search terms.
[0112] In some examples, such as Figure 3 As shown in the figure, this embodiment of the invention also provides a method for automatically completing search terms, and the specific implementation steps of the method are as follows:
[0113] Step 300: First, segment the title of the document to be recalled to obtain each title field, extract the keywords of each title field, and determine the set of suggested words corresponding to the document to be recalled based on all the extracted keywords;
[0114] Step 301: Based on each suggestion word in the suggestion word set corresponding to the document to be recalled, establish a trie for the document to be recalled;
[0115] Step 302: Identify the search term entered by the user, and use a prefix matching algorithm to perform prefix matching on the suggested terms in the trie to determine the suggested terms that match the search term;
[0116] Step 303: Determine the weight of the suggested term based on the degree of matching between the suggested term and the search term;
[0117] Step 304: Adjust the weights of the suggested words based on the text vector normalization algorithm to obtain the recall score of the suggested words;
[0118] Step 305: Based on the recall scores of the suggested words, determine the titles of the documents to be recalled corresponding to a preset number of the highest-scoring suggested words;
[0119] Step 306: Sort the documents to be recalled corresponding to the preset number of highest-scoring suggested words, and display the titles of the documents to be recalled in the sorted order.
[0120] In some examples, such as Figure 4 As shown, following the above process, when a user inputs "maca", the titles of the documents to be recalled corresponding to the matching suggested keywords, from high to low, are: "macalon dessert", "macalon cake", "French macalon", "French macalon dessert", and "Italian macalon dessert", which are displayed in the suggestion box below the search box in order, thus automatically completing the user's input "maca".
[0121] Based on the same inventive concept, this embodiment of the invention also provides a device for completing search content. Since this device is the same as the device in the method of this embodiment of the invention, and the principle of the device in solving the problem is similar to that of the method, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.
[0122] like Figure 5 As shown, the device includes a processor 500 and a memory 501. The memory 501 stores programs executable by the processor 500. The processor 500 reads the programs from the memory 501 and performs the following steps:
[0123] From the set of suggested terms corresponding to each document to be recalled, suggest terms that match the search terms entered by the user are determined, wherein the set of suggested terms is used to represent the key information in the title of the corresponding document to be recalled;
[0124] The recall score of the suggested term is determined based on the degree of matching between the suggested term and the search term;
[0125] Based on the recall scores of the suggested terms, determine the titles of the documents to be recalled corresponding to a preset number of the highest-scoring suggested terms, and display the titles.
[0126] In some examples, the processor 500 is specifically configured to determine the set of suggested words corresponding to the document to be recalled in the following manner:
[0127] The title of the document to be recalled is segmented to obtain individual title fields;
[0128] Extract keywords from each title field, and determine the set of suggested words corresponding to the document to be recalled based on all extracted keywords.
[0129] In some examples, the processor 500 is specifically configured to perform:
[0130] Based on each suggestion word in the suggestion word set corresponding to the document to be recalled, a trie is built for the document to be recalled;
[0131] The suggested words in the trie are matched by prefix matching algorithm to determine the suggested words that match the search term.
[0132] In some examples, the processor 500 is specifically configured to perform:
[0133] The weight of the suggested term is determined based on the degree of matching between the suggested term and the search term;
[0134] The weights of the suggested words are adjusted based on the text vector normalization algorithm to obtain the recall score of the suggested words.
[0135] In some examples, processor 500 is specifically configured to execute:
[0136] Based on the recall score of the suggested keywords, the documents to be recalled corresponding to the suggested keywords are sorted.
[0137] The titles of the documents to be recalled are displayed in the order described.
[0138] Based on the same inventive concept, this embodiment of the invention also provides a device for completing search content. Since this device is the same as the device in the method of this embodiment of the invention, and the principle of the device in solving the problem is similar to that of the method, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.
[0139] like Figure 6 As shown, the device includes:
[0140] The suggestion word unit 600 is used to determine suggestion words that match the search terms entered by the user from the suggestion word set corresponding to each document to be recalled, wherein the suggestion word set is used to represent key information in the title of the corresponding document to be recalled;
[0141] The recall score determination unit 601 is used to determine the recall score of the suggested word based on the degree of matching between the matched suggested word and the search word;
[0142] The title display unit 602 is used to determine the titles of the documents to be recalled corresponding to a preset number of the highest-scoring suggested words based on the recall scores of the suggested words, and to display the titles.
[0143] In some examples, the suggestion word determination unit 600 is specifically used to determine the set of suggestion words corresponding to the document to be recalled in the following manner:
[0144] The title of the document to be recalled is segmented to obtain individual title fields;
[0145] Extract keywords from each title field, and determine the set of suggested words corresponding to the document to be recalled based on all extracted keywords.
[0146] In some examples, the determination of suggestion word unit 600 is specifically used for:
[0147] Based on each suggestion word in the suggestion word set corresponding to the document to be recalled, a trie is built for the document to be recalled;
[0148] The suggested words in the trie are matched by prefix matching algorithm to determine the suggested words that match the search term.
[0149] In some examples, the recall score determination unit 601 is specifically used for:
[0150] The weight of the suggested term is determined based on the degree of matching between the suggested term and the search term;
[0151] The weights of the suggested words are adjusted based on the text vector normalization algorithm to obtain the recall score of the suggested words.
[0152] In some examples, the title display determination unit 602 is specifically used for:
[0153] Based on the recall score of the suggested keywords, the documents to be recalled corresponding to the suggested keywords are sorted.
[0154] The titles of the documents to be recalled are displayed in the order described.
[0155] Based on the same inventive concept, embodiments of the present invention also provide a computer storage medium storing a computer program, which, when executed by a processor, performs the following steps:
[0156] From the set of suggested terms corresponding to each document to be recalled, suggest terms that match the search terms entered by the user are determined, wherein the set of suggested terms is used to represent the key information in the title of the corresponding document to be recalled;
[0157] The recall score of the suggested term is determined based on the degree of matching between the suggested term and the search term;
[0158] Based on the recall scores of the suggested terms, determine the titles of the documents to be recalled corresponding to a preset number of the highest-scoring suggested terms, and display the titles.
[0159] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.
[0160] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 Devices that specify the functions in one or more boxes.
[0161] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including an instruction device, which is implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0162] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0163] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.
Claims
1. A method of completing a search for content, characterized by, The method comprises: corresponding relationship between the document identifier, the document content and the document title of each document to be recalled is constructed, key information of each document to be recalled is extracted, a set of suggestion words is generated by using the key information, and the weight of each suggestion word is set; corresponding relationship between the document identifier, the document content, the document title and the set of suggestion words of each document to be recalled is constructed, and document information to be recalled is generated according to the corresponding relationship; and the document information to be recalled is imported into an index of a search engine, so as to be used for suggestion word matching of a search word input by a user; from the set of suggestion words corresponding to each document to be recalled, a suggestion word matched with the search word input by the user is determined, wherein the set of suggestion words is used to represent key information in the title of the corresponding document to be recalled; the weight of the suggestion word is determined according to the matching degree of the matched suggestion word and the search word; a normalization score of the suggestion word is calculated based on a text vector normalization algorithm; and the product of the weight of the suggestion word and the normalization score is determined as the recall score of the suggestion word; the rule for calculating the normalization score of the suggestion word based on the text vector normalization algorithm is that the quotient of the field length of the search word input by the user divided by the field length of the suggestion word is the normalization score of the suggestion word; the title of the document to be recalled corresponding to the suggestion word with the highest score is determined according to the recall score of the suggestion word, and the title is displayed.
2. The method of claim 1, wherein, The set of suggestion words corresponding to the document to be recalled is determined in the following manner: each title field is obtained by segmenting the title of the document to be recalled; key words of the title fields are extracted, and the set of suggestion words corresponding to the document to be recalled is determined according to all the extracted key words.
3. The method of claim 1, wherein, The set of suggestion words corresponding to the document to be recalled is determined in the following manner: a dictionary tree of the document to be recalled is established based on each suggestion word in the set of suggestion words corresponding to the document to be recalled; prefix matching of the suggestion words in the dictionary tree is performed by using a prefix matching algorithm, and the suggestion word matched with the search word is determined.
4. The method according to any one of claims 1 to 3, characterized in that, The title is displayed in the following manner: the document to be recalled corresponding to the suggestion word is sorted according to the recall score of the suggestion word; the title of the document to be recalled is displayed in the order of the sorting.
5. An apparatus for completing a search for content, the apparatus comprising: The device comprises a processor and a memory, the memory is used to store a program executable by the processor, and the processor is used to read the program in the memory and perform the following steps: corresponding relationship between the document identifier, the document content and the document title of each document to be recalled is constructed, key information of each document to be recalled is extracted, a set of suggestion words is generated by using the key information, and the weight of each suggestion word is set; corresponding relationship between the document identifier, the document content, the document title and the set of suggestion words of each document to be recalled is constructed, and document information to be recalled is generated according to the corresponding relationship; and the document information to be recalled is imported into an index of a search engine, so as to be used for suggestion word matching of a search word input by a user; determining a suggestion word matching the search word input by the user from a suggestion word set corresponding to each document to be recalled, wherein the suggestion word set is used to represent key information in a title of the corresponding document to be recalled; determining a weight of the suggestion word according to a matching degree between the matching suggestion word and the search word; calculating a normalization score of the suggestion word based on a text vector normalization algorithm; determining a recall score of the suggestion word as a product of the weight of the suggestion word and the normalization score; a rule for calculating the normalization score of the suggestion word based on the text vector normalization algorithm is that a quotient of a field length of the search word input by the user divided by a field length of the suggestion word is the normalization score of the suggestion word; determining a title of the document to be recalled corresponding to a preset number of suggestion words with the highest scores according to the recall scores of the suggestion words, and displaying the title.
6. The apparatus of claim 5, wherein, The processor is specifically configured to determine the suggestion word set corresponding to the document to be recalled by: segmenting the title of the document to be recalled to obtain each title field; extracting key words of the each title field, and determining the suggestion word set corresponding to the document to be recalled according to all the extracted key words.
7. The apparatus of claim 5, wherein, The processor is specifically configured to perform: establishing a dictionary tree of the document to be recalled based on each suggestion word in the suggestion word set corresponding to the document to be recalled; determining the suggestion word matching the search word by prefix matching algorithm on the suggestion word in the dictionary tree.
8. The apparatus of any one of claims 5-7, wherein, The processor is specifically configured to perform: sorting the document to be recalled corresponding to the suggestion word according to the recall score of the suggestion word; displaying the title of the document to be recalled in the order of the sorting.