KR20140114496A - Method and system for providing summery of text document using word cloud - Google Patents

Abstract

Disclosed are a method for providing the summary of a word cloud-based text document in an image form and an information providing system using the same. The document summary method may include a step of generating a word cloud by using words in a given document; and a step of providing the word cloud as an image, which is summary information of the document that has been visualized.

Description

TECHNICAL FIELD [0001] The present invention relates to a method of summarizing an image of a word cloud-based text document, and an information providing system using the same.

Embodiments of the present invention are directed to techniques for summarizing text-based documents using Word Cloud.

There is a problem that it is difficult for a user to determine which document to read in a short time among a vast amount of text-based documents (e.g., lecture materials, articles, etc.)

Although it is possible to know the approximate theme by meta information such as the keyword or title of document, it is also difficult to know what contents are actually covered in the text because it requires human effort. In the absence of visual cognitive ability for text-based documents, it is somewhat difficult to predict the text of a text with a small screen, especially in a mobile environment.

Recently, the Internet information retrieval model adds the concept of automatic document summarization to information retrieval through the Internet. For example, when a user searches for information, he presents a category of the document, extracts a part of the document, presents a portion specified in the description, presents a combination of sentences containing the search word, A preview function for extracting and presenting only the text of the document can be provided.

Korean Patent Publication No. 10-2000-0050225 (published on Aug. 05, 2000, an Internet information retrieval system and method based on automatic document summarization) summarizes a main word and a topic sentence by adding an automatic document summarizing concept to Internet information retrieval And the like.

1, a title 101, a snippet 102, an introduction 103, and an abstract 104 are included in the text-based document 100 in addition to the main text, Keywords (or tags) 105, and the like.

When summarizing a document using meta information such as a title (101) or a keyword (105) as one of the automatic document summarizing methods, it is difficult to grasp the contents of the text, need.

In addition, when summarizing a document using introduction (103) or abstract (104), it is easy to grasp the contents of the text because the summary contents are rather long,

Finally, when a document is summarized using a snippet (102) or the like, a document summary can be automatically generated because an algorithm for extracting an important part of the text is used, but the performance of the extraction algorithm is low.

The present invention provides a document summarizing method and an information providing system using the same, which can summarize a text based document into an image that is effective for cognitive ability.

A document summary method capable of summarizing a text-based document using a word cloud and an information providing system using the same are provided.

According to an embodiment of the present invention, a document summary method comprises the steps of: generating, for a given document, a word cloud using words appearing in the document; And providing the word cloud as a visualized image of the summarized information of the document.

According to an aspect, the step of generating the word cloud includes: calculating a frequency of words appearing in the document; And determining the size of each word according to the frequency of the word.

According to another aspect, the step of generating the word cloud comprises: calculating a frequency of words appearing in the document; Determining a size and an arrangement position of each word according to the frequency of the word; And determining a character color of each word according to the size of the word.

According to another aspect, the step of generating the word cloud includes: calculating a frequency of words appearing in the document; Summing the frequency of the words to obtain the total document length of the document; And determining an image size of the word cloud using the total document length of the document.

According to another aspect, a document summarizing method includes: extracting N topics as a set of similar words for an entire document composed of a plurality of documents; And generating a topic cloud, which is a word cloud for the topic, using the similarity for each topic.

According to another aspect, the topic cloud may be provided in a category for classifying the document.

According to another aspect, the color of the font of the topic cloud may be determined for each topic.

According to another aspect, a document summarizing method includes: extracting N topics, which are sets of similar words, over an entire document consisting of a plurality of documents; And determining a font color of the topic cloud for each topic. At this time, in the step of generating the word cloud, the word cloud of the corresponding topic among the N topics may be applied to each word of one document to generate the word cloud.

According to another aspect of the present invention, the step of extracting N items of the set of the similar words can extract the N topics based on the similarity between words using an LDA (Latent Dirichlet Allocation) algorithm.

According to another aspect of the present invention, in the step of generating the word cloud, the words classified into the same topic are determined to have the same letter color for one document, and the words are classified into the same letter color according to the frequency of words appearing in the document. Transparency can be determined.

According to an embodiment of the present invention, a document summarizing method comprises the steps of: extracting N topics, which are sets of similar words, over an entire document consisting of a plurality of individual documents; Determining a text color for each topic for each topic; And generating a word cloud using words appearing in the individual document for each of the individual documents. At this time, the step of generating the word cloud may include determining, for each word of the individual document, a character size according to the frequency of words appearing in the individual document, and applying a color of the corresponding topic among the N topics .

According to an embodiment of the present invention, an information providing system includes: a database storing a plurality of documents and summary information of each document; And a providing unit for providing the summary information with respect to at least one document corresponding to a search word among the plurality of documents, wherein the summary information includes information that summarized information of the document is a visualized image, It can be a word cloud composed of emerging words.

In accordance with an embodiment of the present invention, textual content of a document can be recognized more quickly and efficiently by automatically summarizing and visualizing a text-based document through imaging using a word cloud.

According to the embodiment of the present invention, the summary information on the text-based document can be clustered among the documents related to the topic semantically, thereby improving the quality of the search service and providing the user with the convenience of searching.

According to an embodiment of the present invention, it is possible to provide a search environment of a cognitive / visual form by providing a document search result through an image with respect to a text-based document. In particular, .

1 is an exemplary diagram for illustrating information used to summarize a text-based document.
2 illustrates a flowchart of a document summary method using word cloud in one embodiment of the present invention.
3 is an exemplary diagram for explaining a process of generating a word cloud as summarized information of a document in an embodiment of the present invention.
4 is an exemplary diagram for explaining a word cloud for a topic extracted from an entire document, in an embodiment of the present invention.
5 is an exemplary diagram illustrating a search result in which, in an embodiment of the present invention, a word cloud is provided as summary information of a document.
6 is a block diagram illustrating an internal configuration of an information providing system for providing a word cloud as a summary information of a document in an embodiment of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

The present embodiments summarize a text-based document and provide corresponding summary information, which can be applied to a search system that finds information matching a predetermined condition among data on the Internet.

In general, there is a difference in cognitive ability between text and image between human and machine. In other words, a machine is much faster to interpret text than an image per pixel, and is more efficient in understanding the meaning, whereas humans are quicker and more efficient in recognizing that information through visual images than text in a dictionary.

Accordingly, the present embodiment proposes a technique of summarizing and textizing the text of a text-based document using a word cloud.

In this specification, 'word cloud' may mean a kind of image in which words appearing in a document are graphically displayed in a cloud shape.

2 illustrates a flowchart of a document summary method using word cloud in one embodiment of the present invention. The document summarizing method according to one embodiment can be performed by the information providing system to be described below.

In step S210, the information providing system may generate a word cloud using a word appearing in the document for a given document. Here, Word Cloud is a graphical image concept that determines the size of words using the frequency of words in a document and rearranges them in a cloud form.

In step S220, the information providing system may provide the word cloud generated for the document as summary information of the document. In other words, the information providing system can use the word cloud as the visualized image of the summarized information of the document. For example, when providing the search result corresponding to the search word, the word cloud is compared with the summary information of the document included in the search result .

The process of generating a word cloud will be described in detail as follows.

For example, the information providing system can convert each word into a vector form based on the frequency with which words appear in the document when a text-based document is input. For example, for a document of <apple banana apple fruit apple banana>, it can be converted into a vector value representing frequency of each word such as {apple: 3, banana: 2, fruit: 1} 3, the information providing system can generate the word cloud 302 by scaling and displaying the words appearing in the document 301 according to the size of each word. For example, if the vector value representing the frequency of each word is {apple: 17, banana: 9, fruit: 3}, and assuming that the logarithmic scaling value is 2 (= LOG (2)), [apple: 4, banana: 3, fruit: 1]. And, the information providing system can place the center of the cloud as the scaled value for the words appearing in the document. At this time, the angle at which each word is arranged may be randomly determined within a certain range (e.g., +/- 20 degrees). In addition, the information providing system can determine and apply the character color of each word according to the scaled value in the word cloud. For example, in the case of a word having the largest scaled numerical value, the character color may be differently applied according to the size of each word in the order of red, orange, brown, green, and blue. As described above, in the present embodiment, the size of each word can be determined according to the frequency appearing in the document in generating the word cloud, and further, at least one of the arrangement position and the character color of each word can be determined.

As another example, the information providing system may determine the image size of the word cloud according to the overall length of the document. To this end, the information providing system can calculate the total document length of the document by calculating the frequency of the words appearing in the document and summing the frequency of the words. The information providing system can adjust the log scaling value of the word cloud in proportion to the total document length of the document. For example, the document size may be scaled for LOG 10 if the document is long, and scaled for LOG (2) if the document is short.

As another example, an information providing system can cluster documents in which semantically related subjects are related to an entire document composed of a plurality of documents. For example, the information providing system can extract N topics, which are sets of similar words, which are semantically similar words, based on the similarity between words using LDA (Latent Dirichlet Allocation) algorithm. At this time, the number N of the topics to be extracted from the entire document can be arbitrarily determined. In the present embodiment, a topic refers to a set of words having a specific topic. As shown in FIG. 4, a word cloud (hereinafter referred to as a topic cloud) for each topic is generated in the same manner as described above, Can be expressed. At this time, the information providing system can determine the font color of each topic cloud for each topic with respect to N topics. In other words, the information providing system can classify the words classified in the same topic into the same color, and configure topic clouds having different topics to be words of different letter colors. The information providing system can use the topic cloud as a category for classifying documents, and can provide the user with N topic clouds to specify the subject of the document to be searched by the user.

As another example, the information providing system can extract the N topics from the entire document composed of a plurality of documents through the LDA algorithm, and then determine the font color of the topic cloud for each topic. At this time, the information providing system may select the topic or the most similar topic among the N topics for each word appearing in one individual document, and set the text color of the corresponding word to the color of the selected topic. That is, the information providing system can generate the word cloud of the document by applying the same color as the words classified into the same topic for each word of the document. Furthermore, the information providing system can apply the same letter color to the words classified as the same topic for one individual document, but can determine the transparency of the letter color according to the frequency of the words appearing in the document. For example, the word "dark red" may be assigned to a word appearing ten times in a word corresponding to a topic designated as "red color", and "light red" may be given to a word appearing twice.

As described above, in the present embodiment, a word cloud displayed at a central portion of a text-based document having a high frequency of appearing frequently in a document is generated, and then the generated word cloud is used as a visualized image of the document A text-based document can be summarized and imaged.

As shown in FIG. 5, the information providing system provides summary information of each document to at least one document corresponding to a search word input by a user among a plurality of documents, The word cloud 502 of each document can be provided as summary information of the document. Furthermore, the information providing system can provide a document search function through an image search for a search word, and at this time, the documents imaged by the word cloud can be managed as an image search object together with actual images. Referring to FIG. 5, the word cloud 502 is included in the search result as the summary information of the document corresponding to the specific search word, and the actual images 503 may be included in the search result.

6 is a block diagram illustrating an internal configuration of an information providing system for providing a word cloud as a summary information of a document in an embodiment of the present invention.

As shown in FIG. 6, the information providing system according to an embodiment includes a generating unit 610 for generating summary information of each document, an information providing unit 610 for providing summary information of the document to at least one document corresponding to the search word, A study 620, and a database (not shown) that stores a plurality of documents and summary information of each document.

The generation unit 610 plays a role of generating a word cloud using a word appearing in a document with respect to a given document. For example, when one text-based document is input, the generation unit 610 determines the size of each word on the basis of the frequency with which the words appear in the document, scales the data according to the size of each word, Of the word cloud. At this time, the generator 610 can arrange the words in the middle of the cloud as the scaled numerical values of the words appearing in the document are higher, and the angles at which the words are arranged are randomly set within a certain range (e.g., +/- 20 degrees) You can decide. In addition, the generator 610 can apply a different color to the text according to the size of each word determined based on the frequency of appearing in the document. Then, the generator 610 calculates the frequency of the words appearing in the document to calculate the total document length of the document in such a manner that the frequencies of the words are all summed. Then, the generator 610 calculates the image size of the word cloud Can be determined.

Furthermore, the generating unit 610 may apply the entire document composed of a plurality of documents to the LDA algorithm to extract N words, which are sets of semantically similar similar words, based on the similarity between words. The generating unit 610 may generate a topic cloud for each topic using a similar word classified into the same topic, and may determine the color of each topic cloud for each topic. At this time, the topic cloud may be used as a category for classifying the document, and the topic cloud selected by the user may specify the subject of the document to be searched. In particular, the generator 610 may generate a word cloud of the corresponding individual document by applying the color of the corresponding topic among the N topics for each word of an individual document. In addition, the generator 610 may apply the same letter color to the words classified into the same topic for one individual document, and determine the transparency of the letter color according to the frequency of the words appearing in the document.

The generating unit 610 for performing the above operation may be configured only by a word cloud generator 612 for generating a word cloud for a single document or by grouping the entire document together with the word cloud generator 612 into subject- A topic cloud generator 611 for generating a topic cloud based on the LDA.

The provider 620 may provide the word cloud generated for each document as a summary information of the document through a database search. In other words, the providing unit 620 may use the word cloud as an image in which summarized information of the document is visualized. For example, when providing a search result corresponding to a search word, It can be provided as summary information. In addition, the providing unit 620 may provide a document search function through an image search for a search word. At this time, the documents imaged by the word cloud may be managed as an image search target together with actual images.

As described above, according to the embodiment of the present invention, the text contents of the document can be recognized more quickly and efficiently by automatically summarizing and visualizing the text-based document through image formation using the word cloud. In addition, according to the embodiment of the present invention, summary information about a text-based document can be clustered among documents related to a topic, thereby improving the quality of the search service and providing the user with the convenience of searching. According to an embodiment of the present invention, a cognitive / visual search environment can be provided by providing a document search result through an image with respect to a text-based document. In particular, It can be understood quickly.

The apparatus described above may be implemented as a hardware component, a software component, and / or a combination of hardware components and software components. For example, the apparatus and components described in the embodiments may be implemented within a computer system, such as, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable array (FPA) A programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions. The processing device may execute an operating system (OS) and one or more software applications running on the operating system. The processing device may also access, store, manipulate, process, and generate data in response to execution of the software. For ease of understanding, the processing apparatus may be described as being used singly, but those skilled in the art will recognize that the processing apparatus may have a plurality of processing elements and / As shown in FIG. For example, the processing unit may comprise a plurality of processors or one processor and one controller. Other processing configurations are also possible, such as a parallel processor.

The software may include a computer program, code, instructions, or a combination of one or more of the foregoing, and may be configured to configure the processing device to operate as desired or to process it collectively or collectively Device can be commanded. The software and / or data may be in the form of any type of machine, component, physical device, virtual equipment, computer storage media, or device , Or may be permanently or temporarily embodied in a transmitted signal wave. The software may be distributed over a networked computer system and stored or executed in a distributed manner. The software and data may be stored on one or more computer readable recording media.

The method according to an embodiment may be implemented in the form of a program command that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, and the like, alone or in combination. The program instructions to be recorded on the medium may be those specially designed and configured for the embodiments or may be available to those skilled in the art of computer software. Examples of computer-readable media include magnetic media such as hard disks, floppy disks and magnetic tape; optical media such as CD-ROMs and DVDs; magnetic media such as floppy disks; Magneto-optical media, and hardware devices specifically configured to store and execute program instructions such as ROM, RAM, flash memory, and the like. Examples of program instructions include machine language code such as those produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments. For example, it is to be understood that the techniques described may be performed in a different order than the described methods, and / or that components of the described systems, structures, devices, circuits, Lt; / RTI &gt; or equivalents, even if it is replaced or replaced.

Therefore, other implementations, other embodiments, and equivalents to the claims are also within the scope of the following claims.

610:
620: Offering

Claims (18)

Generating, for a given document, a word cloud using words appearing in the document; And
Providing summarized information of the document as a visualized image to provide the word cloud
&Lt; / RTI &gt; The method according to claim 1,
The step of generating the word cloud comprises:
Calculating a frequency of words appearing in the document; And
Determining a size of each word according to the frequency of the word;
&Lt; / RTI &gt; The method according to claim 1,
The step of generating the word cloud comprises:
Calculating a frequency of words appearing in the document;
Determining a size and an arrangement position of each word according to the frequency of the word; And
Determining a character color of each word according to the size of the word
&Lt; / RTI &gt; The method according to claim 1,
The step of generating the word cloud comprises:
Calculating a frequency of words appearing in the document;
Summing the frequency of the words to obtain the total document length of the document; And
Determining an image size of the word cloud using the total document length of the document
&Lt; / RTI &gt; The method according to claim 1,
Extracting N topics as a set of similar words for an entire document composed of a plurality of documents; And
Generating a topic cloud that is a word cloud for the topic using the similarity for each topic
&Lt; / RTI &gt; 6. The method of claim 5,
The topic cloud being provided as a category for classifying the document
/ RTI &gt; 6. The method of claim 5,
The font color of the topic cloud is determined for each topic
/ RTI &gt; The method according to claim 1,
Extracting N topics, which are sets of similar words, with respect to an entire document composed of a plurality of documents; And
Determining a font color of the topic cloud for each topic
Further comprising:
The step of generating the word cloud comprises:
For each word of one document, generating the word cloud by applying the color of the corresponding topic among the N topics
/ RTI &gt; 9. The method of claim 8,
Wherein the step of extracting N topics, which are sets of the similar words,
Extracting the N topics based on similarity between words using LDA (Latent Dirichlet Allocation) algorithm
/ RTI &gt; 9. The method of claim 8,
The step of generating the word cloud comprises:
For one document, the words classified into the same topic are determined to have the same letter color, but the transparency of the determined letter color is determined according to the frequency of words appearing in the document
/ RTI &gt; Extracting N topics, which are sets of similar words, with respect to an entire document composed of a plurality of individual documents;
Determining a text color for each topic for each topic; And
Generating a word cloud using words appearing in the individual document for each of the individual documents;
Lt; / RTI &gt;
The step of generating the word cloud comprises:
Determining, for each word of the individual document, a character size according to the frequency of words appearing in the individual document, and applying the color of the corresponding topic among the N topics
/ RTI &gt; A database storing a plurality of documents and summary information of each document; And
For providing at least one document corresponding to a search word among the plurality of documents,
Lt; / RTI &gt;
Wherein the summary information is a word cloud consisting of words appearing in the document as summarized information of the document as a visualized image
And an information providing system. 13. The method of claim 12,
A generation unit for generating the word cloud based on the frequency of words appearing in the document,
Further comprising: 13. The method of claim 12,
A generation unit for generating the word cloud based on the frequency of words appearing in the document,
Further comprising:
Wherein the generation unit comprises:
Determining the size of each word according to the frequency of the word
And an information providing system. 13. The method of claim 12,
A generation unit for generating the word cloud based on the frequency of words appearing in the document,
Further comprising:
Wherein the generation unit comprises:
Determining a total document length of the document by summing the frequency of words appearing in the document, and determining an image size of the word cloud using the total document length of the document
And an information providing system. 13. The method of claim 12,
A generation unit for generating the word cloud based on the frequency of words appearing in the document,
Further comprising:
Wherein the generation unit comprises:
Extracting N topics, which are sets of similar words, from the entire document made up of the plurality of documents, determining the color of the text of the topic cloud for each topic, and then, for each word of one document, Generating the word cloud by applying the font color of the topic
And an information providing system. 17. The method of claim 16,
Wherein the generation unit comprises:
Extracting the N topics based on similarity between words using LDA (Latent Dirichlet Allocation) algorithm
And an information providing system. 17. The method of claim 16,
Wherein the generation unit comprises:
For one document, the words classified into the same topic are determined to have the same letter color, but the transparency of the determined letter color is determined according to the frequency of words appearing in the document
And an information providing system.

Images