TWI490704B - Related vocabulary generation system and method - Google Patents

Description

Related vocabulary generation system and method

The present invention relates to a vocabulary production system and is a related vocabulary production system.

In the process of data query, the keyword can determine the accuracy of the search and the completeness of the data. Therefore, multiple related words are usually used for searching. At this time, the user's text acuity and the documents in various fields must be relied upon. Familiarity can only use efficient keywords, and in general, keywords are easily restricted to specific areas and specific languages, and can not obtain more broad and relevant information, resulting in research material restrictions, so Relevant vocabulary with high relevance and reliability, which can help in data search.

At present, the search of domestic databases, such as doctoral thesis or patent database, only provides automatic conversion query and multi-keyword query function for Chinese-English translation of keywords, but does not provide related vocabulary conversion query, so users Rely on personal experience to get a variety of relevant keywords.

The present invention is a related vocabulary generating system and method that can provide related vocabulary with good reliability.

Therefore, the present invention provides a related vocabulary generation system including an operation interface for inputting a keyword sink, a file database for storing a plurality of file files, and a data analysis for normalizing data stored in the file database. a module, a vocabulary analysis module, and a correlation module, the vocabulary analysis module can perform vocabulary comparison and importance evaluation calculation according to the data output by the data analysis module and the keyword, and the correlation module The correlation and the judgment operation may be performed, and the data output by the vocabulary analysis module is calculated and correlated with the keyword, and the operation interface is driven to display a vocabulary whose correlation is greater than or equal to a predetermined value.

The vocabulary analysis module of the above related vocabulary generation system has a synonymous comparison unit, and the synonymous comparison unit judges the vocabulary synonymous with the keyword according to a synonym vocabulary.

The vocabulary analysis module of the related vocabulary generation system has a word frequency inverse ratio file frequency unit and a word frequency unit, and the word frequency inverse ratio file frequency unit uses a word frequency inverse ratio file frequency algorithm to calculate a vocabulary importance degree, and the word frequency unit calculates a vocabulary according to a vocabulary occurrence frequency. Importance.

The correlation module of the related vocabulary generation system has a temporary storage unit capable of storing data, and an ex ante probability operation unit, which can calculate and correlate the vocabulary occurrences.

The vocabulary analysis module of the above related vocabulary generation system also has a natural vocabulary recognition unit that can recognize commonly used modifiers in natural language.

The document database of the above related vocabulary generation system has a patent data unit for storing a patent document, a journal data unit for storing a journal file, and a paper data unit for storing a doctoral thesis.

In addition, the present invention provides a related vocabulary generating method including the following steps: a data parsing step, a normalized document data; a keyword step, inputting a keyword sink; a synonymous comparison step, comparing at least one keyword with the keyword according to a synonym vocabulary Synonymous vocabulary; important evaluation steps, based on the data obtained by the data analysis step, calculate the importance of each vocabulary; correlation steps, based on the data analysis step and the information obtained by the important evaluation step, statistics and calculation of each vocabulary and the key The relevance of the vocabulary; and the list display step, showing a vocabulary that is synonymous with the keyword and whose correlation is greater than a predetermined value.

The specific features and effects of the present invention are: the synonymous comparison unit can obtain the synonymous vocabulary of the keyword sink in a comparison manner, and after the word frequency unit and the word frequency inverse ratio file frequency unit can determine the importance degree and the discrimination degree of each vocabulary, The correlation module further calculates the probability of the interaction between each vocabulary and the keyword to evaluate the probability of occurrence of each vocabulary and the keyword, to obtain the relevance reordering, thereby generating relevant vocabulary as a standard. Improve the reliability of related vocabulary.

(10) ‧‧‧Related vocabulary production system

(2) ‧‧‧Operator interface

(3) ‧ ‧ document database

(31) ‧ ‧ Patent Data Unit

(32) ‧‧‧Journal data unit

(33) ‧ ‧ paper data unit

(34) ‧ ‧ magazine data unit

(35) ‧‧‧ Seminar data unit

(4) ‧‧‧Data Analysis Module

(5) ‧ ‧ vocabulary analysis module

(51)‧‧‧Synonymous comparison unit

(52) ‧ ‧ word frequency inverse ratio file frequency unit

(53)‧‧‧Word frequency unit

(54)‧‧‧Natural vocabulary identification unit

(6) ‧‧‧ Relevance modules

(61) ‧‧‧ temporary storage unit

(62) ‧‧‧Pre-probability arithmetic unit

(71)‧‧‧Keyword steps

(72) ‧‧‧ Data analysis steps

(73) ‧‧‧ Synonymous control steps

(74) ‧ ‧ important assessment steps

(75) ‧‧‧ Relevance steps

(76) ‧‧‧List display steps

(77) ‧‧‧ association diagram steps

The first block diagram shows the vocabulary generation system of the present invention; the second diagram is a schematic diagram of the screen, which mainly displays an operation interface, and inputs a keyword and displays related words in a list; The third diagram is a schematic diagram of the screen, mainly showing the operation interface and displaying the related vocabulary in a graphical manner; the fourth diagram is a flowchart showing the vocabulary generation method of the present invention; and the fifth diagram is a schematic diagram showing the operation. Interface, and set the vocabulary threshold threshold.

As shown in the first to fourth figures, the present invention relates to a related vocabulary generation system (10), an operation interface (2) for displaying a picture and inputting a keyword, and a file database for storing a plurality of file files (3) A data analysis module (4), a vocabulary analysis module (5), and a correlation module (6) for normalizing the data stored in the document database (3).

The document database (3) has a patent data unit (31), a journal data unit (32), a paper data unit (33), a magazine data unit (34), and a seminar data unit (35). And they are used to store the archives of patent documents, periodicals, master's doctoral thesis, magazines and seminar papers, and can be formalized through the data analysis module (4).

The vocabulary analysis module (5) can perform vocabulary comparison and importance evaluation calculation according to the data output by the data analysis module (4), and has a natural vocabulary identification unit (54) and a synonym comparison unit. (51), the word frequency inverse ratio file frequency unit (52), and a word frequency unit (53). The natural vocabulary recognition unit (54) can identify and filter commonly used modifiers in natural language, such as articles and quantifiers. The synonymous comparison unit (51) is determined based on a plurality of synonym vocabulary comparisons to obtain a vocabulary that is synonymous with the keyword. In the preferred embodiment, the synonym vocabulary is a Chinese-English comparison .

The word frequency unit (53) and the word frequency inverse ratio file frequency unit (52) are mainly used to calculate and evaluate the importance of each vocabulary, and the word frequency inverse ratio file frequency unit (52) is a word frequency inverse ratio file frequency method (TF/IDF). Evaluate the importance of vocabulary, that is, the more important the vocabulary appears to be focused on a particular document, and the word frequency unit (53) evaluates the vocabulary importance based on the frequency of vocabulary occurrence, that is, the importance is positively correlated with the frequency of occurrence. . Further, according to the evaluation result, the vocabulary whose importance is not reached is excluded, and the user can set the threshold value of the vocabulary occurrence frequency (as shown in the fifth figure).

The correlation module (6) can perform correlation and judgment operations, which are mainly used for calculating and judging the relevance of each vocabulary to the keyword, and the synonymous vocabulary has a temporary storage unit (61) capable of storing data. And an event probability unit (62). The ex ante probability unit (62) can calculate and calculate the correlation between the vocabulary occurrences. Specifically, the Apriori algorithm is used to evaluate the probability of occurrence of another vocabulary in the case of a specific vocabulary. This evaluates the correlation between the vocabulary and further compares it with a predetermined value, and displays a vocabulary that is more than or equal to the predetermined value and synonymous with the keyword. The Apriori algorithm is a commonly used and well-known method for mining Association Rules (AR), and is not a feature of the present invention, and therefore will not be further described.

The related vocabulary generating method can be executed by the related vocabulary generating system (10), and includes the following steps: performing a keyword step (71): the user inputs a keyword sink in the operation interface (2), and is preferred In the embodiment, "ERP" is input.

Perform data analysis step (72): the data analysis module (4) is to the file database (3) The stored file file performs normalization operations.

Next, a synonymous comparison step (73) is performed: the natural vocabulary recognition unit (54) of the vocabulary analysis module (5) identifies and filters commonly used modifiers in the natural language, and the synonymous comparison unit (51) is based on a synonym vocabulary In the preferred embodiment, the words "Enterprise Resource Planning" and "Enterprise Resource Planning" are synonymous with "ERP", that is, related words.

In addition to the synonymous related vocabulary, a vocabulary related to the keyword faucet can be obtained, and an important evaluation step (74) is performed: the vocabulary analysis module (5) obtains the data analytic step (72). Data, calculate the importance of each vocabulary, which can be calculated using the word frequency or word frequency inverse file frequency method. The word frequency unit (53) counts the frequency of occurrence of each vocabulary, and its importance is positively correlated with the frequency of occurrence. In addition, the word frequency inverse ratio file frequency unit (52) uses the word frequency inverse ratio file frequency method to evaluate the vocabulary importance degree, and eliminates the vocabulary whose importance is not up to the threshold, specifically, the vocabulary with low frequency of occurrence.

And then performing a correlation step (75): the correlation module (6) calculates and correlates the vocabulary with the keyword according to the data obtained by the data analysis step (72) and the important evaluation step (74), Specifically, the pre- chance calculation unit (62) of the correlation module (6) uses the pre-existing probability calculation to quantify the correlation between each vocabulary and the keyword, and then sorts according to the relevance data.

A list display step (76) is performed: the operation interface (2) displays a vocabulary that is synonymous with the keyword and whose correlation is greater than the predetermined value, that is, the related vocabulary of the keyword sink is displayed in a list manner.

On the other hand, the related vocabulary can also be presented in a schema manner, and the association graph step (77) is performed: the operation interface (2) displays a vocabulary that is synonymous with the keyword and has a correlation greater than a predetermined value, and is based on the correlation strength. Displayed in different colors. Specifically, the display status is: the keyword is located in the center of the screen, and the related vocabulary synonymous with the keyword is displayed at the bottom of the dark frame, and the keyword is connected by a line respectively, and the keyword is highly correlated. The vocabulary is displayed at the bottom of the light-colored frame and connected to the keyword by a line, while the less relevant vocabulary is presented in black on a white background and is not connected to the keyword.

In addition, in the important evaluation step (74), the user can select the source of the data to be evaluated, that is, one of the patent documents, papers, journals or magazines can be selected as the basis for the evaluation, and can be based on The domain that the user actually uses to produce relevant vocabulary that is closer to the needs.

The synonymous comparison unit (51) of the related vocabulary generation system (10) of the present embodiment can obtain the synonymous vocabulary of the keyword sink in a comparison manner, and can judge the word frequency unit (53) and the word frequency inverse ratio file frequency unit (52). After the importance and discrimination of each vocabulary, the correlation module (6) further calculates the probability of each vocabulary and the keyword convergence to evaluate the probability that each vocabulary is accompanied by the keyword to obtain the vocabulary. Correlation reordering, that is, obtaining the correlation between each vocabulary and the keyword, generates relevant vocabulary as a standard, thereby improving the reliability of the related vocabulary.

On the other hand, in addition to presenting relevant vocabulary in a list, the operation interface (2) can also graphically present the relevance of vocabulary, and at the same time, the depth of the color is correlated, and the user can identify the vocabulary and The effect of the keyword exchange relationship.

However, the above description is only a preferred embodiment of the present invention and is not intended to be limiting. The features of the present invention are still within the scope of the present invention for the re-creation of the technical means and the creation principle using the invention. Therefore, the statements of the present invention and the description of the present invention are not intended to limit the scope of the present invention; any changes and modifications made by those skilled in the art without departing from the spirit and scope of the invention are intended to be included in the invention. Within the scope of patents, it is reasonable.

(10) ‧‧‧Related vocabulary production system

(2) ‧‧‧Operator interface

(3) ‧ ‧ document database

(31) ‧ ‧ Patent Data Unit

(32) ‧‧‧Journal data unit

(33) ‧ ‧ paper data unit

(34) ‧ ‧ magazine data unit

(35) ‧‧‧ Seminar data unit

(4) ‧‧‧Data Analysis Module

(5) ‧ ‧ vocabulary analysis module

(51)‧‧‧Synonymous comparison unit

(52) ‧ ‧ word frequency inverse ratio file frequency unit

(53)‧‧‧Word frequency unit

(54)‧‧‧Natural vocabulary identification unit

(6) ‧‧‧ Relevance modules

(61) ‧‧‧ temporary storage unit

(62) ‧‧‧Pre-probability arithmetic unit

Claims (7)

A related vocabulary generation system includes: an operation interface for inputting a keyword sink; a file database for storing a plurality of file files; and a data analysis module for normalizing the data stored in the file database; The vocabulary analysis module can perform vocabulary comparison and importance evaluation calculation according to the data output by the data analysis module and the keyword, and the lexical analysis module has a synonymous comparison unit, and the synonym control unit is based on at least one different language. a synonym vocabulary for judging a vocabulary synonymous with the keyword, the synonym vocabulary containing at least one Chinese-English comparison; and a correlation module for performing correlation and judgment operations, and calculating the output of the lexical analysis module The data is correlated with the keyword, and drives the operation interface to display a vocabulary whose correlation is greater than or equal to a predetermined value. The correlation module has a temporary storage unit capable of storing data, and an ex ante probability unit, the probability of the event The arithmetic unit can calculate and correlate the companion between vocabulary and use the Apriori algorithm to evaluate In the case where a specific vocabulary appears, another vocabulary is accompanied by the probability of occurrence, thereby evaluating the correlation between vocabulary and further comparing with a predetermined value, and displaying a vocabulary having a correlation with the keyword greater than or equal to the predetermined value and synonymous The operation interface displays words that are synonymous with the keyword and whose correlation is greater than a predetermined value, and are displayed in different colors according to the correlation strength, and the keyword is Located in the center of the screen, the relevant vocabulary synonymous with the keyword is displayed at the bottom of the dark frame, and the keyword is connected by a line, and the vocabulary with high relevance is displayed at the bottom of the light frame and connected by lines. Key words sink, while less relevant words are presented in black on white and are not connected to the keyword. For example, the related vocabulary generation system described in claim 1 wherein the vocabulary analysis module has a word frequency inverse ratio file frequency unit and a word frequency unit, and the word frequency inverse ratio file frequency unit uses word frequency inverse ratio file frequency algorithm to calculate vocabulary importance. Degree, the word frequency unit calculates the vocabulary importance according to the frequency of vocabulary occurrence. For example, the related vocabulary generation system described in claim 1 has a patent data unit for storing a patent document, a journal data unit for storing a journal file, and a paper data unit for storing a doctoral thesis. The related vocabulary generation system of claim 1, wherein the vocabulary analysis module further has a natural vocabulary recognition unit that can recognize commonly used modifiers in natural language. A related vocabulary generating method comprises the following steps: a data parsing step, a normalized document data, a keyword step, inputting a keyword sink, and a synonymous comparison step, comparing at least one synonym with the keyword according to a synonym vocabulary of at least one different language. Vocabulary, the synonym vocabulary contains at least one Chinese-English comparison; the important evaluation step, based on the data obtained by the data analysis step, calculates the importance of each vocabulary; The correlation step is based on the data analysis step and the data obtained by the important evaluation step, and the correlation between the vocabulary and the keyword is calculated and calculated, and the correlation probability unit is used to calculate and correlate the collocation between the vocabulary. And using the Apriori algorithm to evaluate the probability of occurrence of another vocabulary in the case of a specific vocabulary, to evaluate the correlation between vocabulary, and further compare with a predetermined value, and display the correlation with the keyword is greater than Or a predetermined value and a synonymous vocabulary; and a list display step of displaying a vocabulary that is synonymous with the keyword and having a correlation greater than a predetermined value; further presenting the related vocabulary in a schema manner, and performing the association graph step: displaying and The keyword is synonymous with a vocabulary that is more than a predetermined value and is displayed in a different color depending on the strength of the correlation. The keyword is located in the center of the screen, and the related vocabulary synonymous with the keyword is displayed at the bottom of the dark frame, and the keyword is connected by a line, and the keyword with high relevance is displayed at the bottom of the light frame and Lines connect to the keyword sink, while less relevant words are presented in black on white and are not connected to the keyword. For example, in the related vocabulary generating method described in claim 5, in the important evaluation step, the importance is calculated according to the vocabulary occurrence, and the importance is proportional to the number of vocabulary occurrences. The related vocabulary generating method according to claim 5, wherein in the important evaluation step, the word frequency inverse ratio file frequency algorithm is used to calculate the vocabulary importance.