A method for related word retrieval based on structured data

Abstract

The present invention discloses a method for searching related words based on structured data, comprising the following steps: S1, setting a threshold of the related level; S2, inputting search words; S3, defining a word related space; S4, dividing the related levels; S5, grouping related words; S6, presenting related words. Compared with the traditional method for searching related words, the present invention provides a method for searching related words based on structured data. After the input word and its related synonyms, antonyms, close words, extended words, homophones, fuzzy-sounding words, related pictures, related videos and related public documents containing the word are graded, the words other than the input word are graded again, and the words are grouped and displayed to the user according to the priority level, which improves the accuracy of the results after the user's search to a certain extent.

Description

Technical Field

[0001] The present invention belongs to the field of related word retrieval, and more specifically, particularly relates to a related word retrieval method based on structured data. Background Art

[0002] Structured data, in simple terms, is a database. It is easier to understand when combined with typical scenarios, such as enterprise ERP, financial systems; medical HIS databases; education cards; government administrative approvals; other core databases, etc. What storage solutions do these applications require? Basically, they include high-speed storage application requirements, data backup requirements, data sharing requirements, and data disaster recovery requirements. With the rapid increase in the amount of information and data on the Internet, search engines need to take auxiliary means to improve search efficiency in order to better serve users. Intelligent recommendation of related words is one of the many auxiliary means.

[0003] At present, most of the existing technologies classify the input word and its related synonyms, near-synonyms, antonyms, similar words, expanded words, homophones, fuzzy-sounding words, related pictures, related videos and related public documents containing the word into levels and display them to users. However, the synonyms, near-synonyms, antonyms, similar words, expanded words, homophones, fuzzy-sounding words, related pictures, related videos and related public documents containing the word of the input word are not divided in detail, which affects the accuracy of user search to a certain extent. Therefore, we propose a related word retrieval method based on structured data. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a related word retrieval method based on structured data.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A method for searching related terms based on structured data comprises the following steps:

[0007] S1. Setting a relevance level threshold, wherein the relevance level threshold is set according to the density of occurrence of the associated words, and the threshold range is ≥5%, that is, the associated words appear at least 5 times among 100 words;

[0008] S2. Enter the search terms, a group of terms or a word set to be searched, and a separator is required between the word sets;

[0009] S3. Define a word association space, where the word association space includes synonyms, near-synonyms, antonyms, similar words, extended words, homophones, fuzzy-sounding words, related pictures, related videos, and related public documents containing the word;

[0010] S4, dividing the association level, respectively based on the density of each target associated word, calculating the association degree between each target associated word and the search word, and dividing the association level according to the association degree, the association level includes five levels: A, B, C, D, and E, and level A is the highest level. The association is divided according to the density of the associated words. The higher the density of the associated words, the higher the association level;

[0011] S5, grouping related words, grouping target words, pictures, videos and documents whose relevance reaches a set threshold as search related words, and grouping them from high to low according to their relevance density, and each group shall not exceed ten target words, pictures, videos and documents whose relevance reaches a set threshold;

[0012] S6. Presentation of related words: presenting the grouped target words, pictures, videos, and documents to the user as a group.

[0013] Preferably, the association space described in step S3 summarizes the degree relationship words of the input words by contrast calculation and deductive calculation, and the degree relationship words include: principal-subordinate relationship words, member relationship words, similarity relationship words, opposition relationship words, and parallel relationship words.

[0014] Preferably, the deductive calculation uses transfer rules, inversion rules and addition and subtraction rules for deductive calculation, and the transfer rules, inversion rules and addition and subtraction rules are specifically:

[0015] 1) Transfer rule, A(b, c), c(d, e), then A(b, c) also exists with A(d, e), where A is the input word, b and c are related words of A, and d and e are related words of c;

[0016] 2) Inversion rule, R(b, c), R(c, d), then A(b, c) and there exists A(c, d), where A is the input word, R is the antonym of A, and both the group b, c and the group c, d are the antonyms of R. Then the inversion results in the group b, c and the group c, d being the synonyms of A;

[0017] 3) Addition and subtraction rules, A(b, c), A(d, e), then A(b, c)+A(d, e)=A(a, ​​b, c, d), where b, c, d, e are all associated words of the input word A.

[0018] Preferably, the synonyms, near synonyms, antonyms, similar words, extended words, homophones, fuzzy-sounding words, related pictures, related videos and related public documents containing the words described in step S4 are estimated by coefficient estimation method, and the association levels are divided completely according to the coefficient values, wherein the coefficient value is calculated as follows:

[0019] D=C(1+I)·k

[0020] Where D is the coefficient value of the associated words, C is the coefficient cardinality average value, I is the average priority level of the group of associated words, and k is the density percentage of the associated words;

[0021] Among them, the priority level of the input word body is the highest level 5, the priority level of the synonyms, near-synonyms, and similar words is 4, the priority level of the extended words, homophones, fuzzy-sounding words, and related public documents containing the words is 3, and the priority level of the antonyms, related pictures, and related videos is 2;

[0022] Among them, the coefficient base is specifically set as: the coefficient base of the input word is 100; the coefficient base of the synonyms, near-synonyms, and similar words of the input word is 80; the coefficient base of the expanded words, homophones, fuzzy-sounding words of the input word, and related public documents containing the word is 50; the coefficient base of the antonym of the input word, related pictures, and related videos is 30.

[0023] Preferably, the association level described in step S4 is specifically A≥100 coefficient value, B≥85 coefficient value, C≥60 coefficient value, D≥40 coefficient value, and E≥25 coefficient value.

[0024] Preferably, the grouping of the associated words in step 5 is specifically based on:

[0025] 1) Group them in order from level A to E;

[0026] 2) When the number of high-level groups is insufficient, the next group will be supplemented, and the coefficients of the supplementary associated words will be filled in order from high to low. For example, when the number of contents in Group A is less than 2, the two contents with the highest coefficients are extracted from Group B to supplement Group A. When the number of contents in Group B is insufficient, the above method is used to extract contents from Group C to supplement, and so on until Group E. When the number of contents in Group E is insufficient, it will not be supplemented;

[0027] 3) Related pictures and related videos are grouped separately, and the related pictures and related videos are not presented simultaneously with the rest of the associated space content.

[0028] Preferably, the input search terms described in step S2 are input through a communication terminal or a retrieval device terminal, and the communication terminal and the retrieval device terminal both include a wireless signal transmitting unit, and the wireless signal transmitting unit is used to convert the input data into an analog signal or a digital signal for transmission, the analog signal is CVBS, S-VIDEO, VGA; the digital signal is ITU-RBT.656, ITU-RBT.601, ITU-RBT.1120, DVI, HDMI.

[0029] Preferably, the separator described in step S1 is specifically a field separator, and the field separator can be one or more of a blank key, a table key, and an enter key. The separator is used to decompose a word command line input through a communication terminal or a search device terminal into fields, and define a word association space according to the decomposed key fields.

[0030] Technical effects and advantages of the present invention: Compared with the traditional related word retrieval method, the present invention provides a related word retrieval method based on structured data. After the input word and its related synonyms, near-synonyms, antonyms, similar words, extended words, homophones, fuzzy-sounding words, related pictures, related videos and related public documents containing the word are graded, the words other than the input word are graded again, and the words are displayed to the user in groups according to priority levels, which improves the accuracy of the results after the user's search to a certain extent. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 The figure is a flow chart of the associated word retrieval method based on structured data of the present invention. DETAILED DESCRIPTION

[0032] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention is further described in detail below in conjunction with specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0033] The present invention provides Figure 1 A method for searching related words based on structured data comprises the following steps:

[0034] S1. Setting a relevance level threshold. The relevance level threshold is set according to the density of the associated words, and the threshold range is ≥5%, that is, the associated words appear at least 5 times among 100 words;

[0035] S2. Enter the search terms, a group of terms or a word set to be searched, and a separator is required between the word sets;

[0036] S3. Define the word association space, which includes synonyms, antonyms, similar words, extended words, homophones, fuzzy-sounding words, related pictures, related videos, and related public documents containing the word;

[0037] S4, dividing the association level, based on the density of each target associated word, calculating the association degree between each target associated word and the search word, and dividing the association level according to the association degree, the association level includes five levels: A, B, C, D, and E, and the A level is the highest level. The association is divided according to the density of the associated words. The higher the density of the associated words, the higher the association level;

[0038] S5. Grouping related words: grouping target words, pictures, videos and documents whose relevance reaches the set threshold as search related words. The groups are grouped from high to low according to their relevance density, and each group does not exceed ten target words, pictures, videos and documents whose relevance reaches the set threshold;

[0039] S6. Presentation of related words: presenting the grouped target words, pictures, videos, and documents to the user as a group.

[0040] The association space of step S3 summarizes the degree relationship words of the input words by contrast calculation and deductive calculation. The degree relationship words include: principal-subordinate relationship words, member relationship words, similarity relationship words, opposition relationship words, and parallel relationship words.

[0041] Deductive calculation uses transfer rules, inversion rules and addition and subtraction rules for deductive calculation. The transfer rules, inversion rules and addition and subtraction rules are as follows:

[0042] 1) Transfer rule, A(b, c), c(d, e), then A(b, c) also exists with A(d, e), where A is the input word, b and c are related words of A, and d and e are related words of c;

[0043] 2) Inversion rule, R(b, c), R(c, d), then A(b, c) and there exists A(c, d), where A is the input word, R is the antonym of A, and both the group b, c and the group c, d are the antonyms of R. Then the inversion results in the group b, c and the group c, d being the synonyms of A;

[0044] 3) Addition and subtraction rules: For A(b, c) and A(d, e), then A(b, c) + A(d, e) = A(a, b, c, d), where b, c, d, and e are all related words of the input word A.

[0045] For the synonyms, near-synonyms, antonyms, related words, extended words, homophones, words with fuzzy sounds, related pictures, related videos, and related published documents containing the word in step S4, the coefficient estimation method is used to estimate their coefficient values, and the association level is completely divided according to the coefficient values. The calculation method of the coefficient value is:

[0046] D = C(1 + I)·k

[0047] Where D is the coefficient value of the related word, C is the average value of the coefficient base, I is the average value of the priority levels of this group of related words, and k is the density percentage of the appearance of the related word;

[0048] Among them, the priority level of the input word ontology is the highest level 5, the priority levels of synonyms, near-synonyms, and related words are level 4, the priority levels of extended words, homophones, words with fuzzy sounds, and related published documents containing the word are level 3, and the priority levels of antonyms, related pictures, and related videos are level 2;

[0049] Among them, the coefficient base is specifically set as follows: the coefficient base of the input word is 100; the coefficient bases of the synonyms, near-synonyms, and related words of the input word are 80; the coefficient bases of the extended words, homophones, words with fuzzy sounds, and related published documents containing the word are 50; the coefficient bases of the antonyms, related pictures, and related videos of the input word are 30.

[0050] Exemplarily, in Example 1, the input word is "add", and the coefficient base of its ontology is 100, and the priority level is level 5. If the density percentage of the appearance of "add" in a certain text is 10% and its related words do not appear, according to the formula

[0051] D = C(1 + I)·k

[0052] The coefficient value of this text is obtained as D = 100×(1 + 5)×10% = 60;

[0053] In Example 2, the input word is "add". If the density percentage of the simultaneous appearance of "add" and its near-synonym "enter" in a certain text is 30%, according to the above formula, the coefficient value of this text is obtained as D = 90×(1 + 4.5)×30% = 148.5;

[0054] Among them, in the above example, the average value of the coefficient bases of the input word "add" and its related word "enter" is (100 + 80) / 2 = 90, and the average value of the priority levels of its related words is (5 + 4) / 2 = 4.5.

[0055] The association level of step S4 is specifically A≥100 coefficient value, B≥85 coefficient value, C≥60 coefficient value, D≥40 coefficient value, and E≥25 coefficient value;

[0056] Exemplarily, by substituting the above-calculated coefficient value into the association level, it is concluded that the association level of the paragraph of text in Example 1 is C level; and the association level of the paragraph of text in Example 2 is A level.

[0057] The specific grouping basis of the associated words in step 5 is:

[0058] 1) Group them in order from level A to E;

[0059] 2) When the number of high-level groups is insufficient, the next group will be supplemented, and the coefficients of the supplementary associated words will be filled in order from high to low. For example, when the number of contents in Group A is less than 2, the two contents with the highest coefficients are extracted from Group B to supplement Group A. When the number of contents in Group B is insufficient, the above method is used to extract contents from Group C to supplement, and so on until Group E. When the number of contents in Group E is insufficient, it will not be supplemented;

[0060] 3) Related pictures and related videos are grouped separately, and the related pictures and related videos are not presented at the same time as the rest of the related space content;

[0061] In the above-mentioned grouping of related words, related pictures and related videos are not grouped together with texts. Their grouping also requires the classification of association levels of their names. The classification rules are the same as the word association level classification method. When classifying the association levels of related pictures and related videos, the higher level between the content association level and the name association level is taken as the final association level.

[0062] The input search terms in step S2 are input through a communication terminal or a search device terminal, and the communication terminal and the search device terminal both include a wireless signal transmitting unit, which is used to convert the input data into an analog signal or a digital signal for transmission. The analog signals are CVBS, S-VIDEO, VGA; the digital signals are ITU-RBT.656, ITU-RBT.601, ITU-RBT.1120, DVI, HDMI.

[0063] The delimiter in step S1 is specifically a field delimiter, which can be one or more of a blank key, a table key, and an enter key. The delimiter is used to decompose the word command line input through a communication terminal or a retrieval device terminal into fields, and define the word association space according to the decomposed key fields.

[0064] To sum up: compared with the traditional related word retrieval method, the related word retrieval method provided by the present invention is a related word retrieval method based on structured data. After the input word and its related synonyms, near-synonyms, antonyms, similar words, extended words, homophones, fuzzy-sounding words, related pictures, related videos and related public documents containing the word are graded, the words other than the input word are graded again, and the words are displayed to the user in groups according to priority levels, which improves the accuracy of the results after the user's search to a certain extent.

[0065] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for searching related terms based on structured data, characterized in that: The steps include: S1. Setting a relevance level threshold, wherein the relevance level threshold is set according to the density of occurrence of the associated words, and the threshold range is ≥5%, that is, the associated words appear at least 5 times among 100 words; S2. Enter the search terms, a group of terms or a word set to be searched, and a separator is required between the word sets; S3. Define a word association space, where the word association space includes synonyms, near-synonyms, antonyms, similar words, extended words, homophones, fuzzy-sounding words, related pictures, related videos, and related public documents containing the search term; S4, dividing the association level, based on the density of each target associated word, calculating the association degree between each target associated word and the search word, and dividing the association level according to its association degree, the association level includes five levels: A, B, C, D, and E, among which level A is the highest level, and the association level is divided according to the density of the associated words. The higher the density of the associated words, the higher the association level; the synonyms, near synonyms, antonyms, similar words, extended words, homophones, fuzzy sound words, related pictures, related videos and related public documents containing the search word in step S4 are estimated by coefficient estimation method, and the association level is divided completely according to the coefficient value, wherein the coefficient value is calculated as follows: Among them, D is the coefficient value of the associated words, C is the average value of the coefficient cardinality, is the average value of the priority level of the associated words, k is the density percentage of the associated words; The priority level of the input search term is level 5, the priority level of synonyms, near synonyms, and similar words is level 4, the priority level of expanded words, homophones, fuzzy words, and related public documents containing the search term is level 3, and the priority level of antonyms, related pictures, and related videos is level 2; The coefficient bases are specifically set as follows: the coefficient base of the input search term is 100; the coefficient base of the synonyms, near synonyms, and similar words of the input search term is 80; the coefficient base of the expanded words, homophones, fuzzy words, and related public documents containing the search term of the input search term is 50; the coefficient base of the antonyms, related pictures, and related videos of the input search term is 30; The association level described in step S4 is specifically A≥100 coefficient value, B≥85 coefficient value, C≥60 coefficient value, D≥40 coefficient value, and E≥25 coefficient value; S5, grouping related words, grouping target words, pictures, videos and documents whose relevance reaches a set threshold as search related words, and grouping them from high to low according to their relevance, and each group shall not exceed ten target words, pictures, videos and documents whose relevance reaches a set threshold; S6. Presentation of related words: presenting the grouped target words, pictures, videos and documents to the user in groups.

2. The method for searching related terms based on structured data according to claim 1, characterized in that: The word association space described in step S3 summarizes the degree relationship words of the input search words by contrast calculation and deductive calculation. The degree relationship words include: main-subordinate relationship words, member relationship words, similarity relationship words, opposition relationship words, and parallel relationship words.

3. The method for searching related terms based on structured data according to claim 2, characterized in that: The deductive calculation uses transfer rules, inversion rules and addition and subtraction rules to perform deductive calculations. The transfer rules, inversion rules and addition and subtraction rules are specifically: 1) Transfer rule, A(b, c), c(d, e), then A(b, c) also exists with A(d, e), where A is the input search term, b and c are related terms of A, and d and e are related terms of c; 2) Inversion rule, R(b, c), R(c, d), then A(b, c) and there exists A(c, d), where A is the input search term, R is the antonym of A, (b, c) and (c, d) are both antonyms of R, then the inversion result is that (b, c) and (c, d) are both synonyms of A; 3) Addition and subtraction rules, A(b, c), A(d, e), then A(b, c)+A(d, e)=A(b, c, d, e), where b, c, d, e are all related terms of the input search term A.

4. The method for searching related terms based on structured data according to claim 1, characterized in that: The specific grouping basis of the associated words grouping in step 5 is: 1) Group them from level A to E; 2) When the number of words in the higher-level group is insufficient, the next group will be used to supplement the words, and the coefficients of the supplementary related words will be filled in order from high to low; 3) Related pictures and related videos are grouped separately, and the related pictures and related videos are not presented simultaneously with the rest of the word-related space content.

5. The method for searching related terms based on structured data according to claim 1, characterized in that: The input search terms described in step S2 are input through a communication terminal or a retrieval device terminal, and the communication terminal and the retrieval device terminal both include a wireless signal transmitting unit, and the wireless signal transmitting unit is used to convert the input data into an analog signal or a digital signal for transmission, the analog signal is CVBS, S-VIDEO, VGA; the digital signal is ITU-RBT.656, ITU-R BT.601, ITU-R BT.1120, DVI, HDMI.

6. The method for searching related terms based on structured data according to claim 5, characterized in that: The separator described in step S2 is specifically a field separator, and the field separator can be one or more of a blank key, a table key, and an enter key. The separator is used to decompose the search terms input through a communication terminal or a search device terminal into fields, and define the term association space according to the decomposed key fields.

Images