KR20170118399A - Method and apparatus of search using big data - Google Patents

Abstract

A method and apparatus for searching keywords using big data through an electronic terminal are provided. And sets a big data database to be searched in the electronic terminal. The search keyword is input through the input unit of the electronic terminal. The electronic terminal selects data including the search keyword from each of the data included in the big data database. The electronic terminal counts the number of words or phrases included in the selected data. The electronic terminal places the counted words or phrases in order of appearance frequency. The electronic terminal selects a plurality of ranked words or phrases as a plurality of related search words. The display of the electronic terminal visually displays a plurality of selected related search terms.

Description

TECHNICAL FIELD [0001] The present invention relates to a method and apparatus for searching large data,

The present invention relates to a searching method and apparatus using big data.

BACKGROUND ART [0002] Recently, information processing technology has rapidly developed and a query has been input from a user in an information processing apparatus such as a smart phone, a tablet PC, a PDA (Personal Digital Assistant), and the like, Researches on search methods and systems that are provided to users are being actively conducted.

Particularly, since the era of Big Data has come, it is possible to quickly search and provide accurate information corresponding to a user query in a plurality of databases in which a large amount of data is recorded through a search engine system, , Intelligent robots, next-generation PCs, telematics, and home networks.

It has been a while since the concept of big data has been around, but the use of such big data has been mainly in the fields of companies dealing with individuals, companies dealing with companies, or in the public sector.

Meanwhile, the existing Internet keyword search searches the data stored in the related server of the portal site or the data on the Internet, for example, and finds and presents the data containing (or related to) the keyword. Most of the materials presented are web pages, picture files, video files, and so on. In many cases, except for pictures, videos, etc., homepage links with text data are often searched.

A conventional search finds specific data related to a search keyword (i.e., a specific press article, a specific blog article, a specific image, a specific video, etc.). In the past, the top most search result (homepage link) may or may not represent 'Hong Gil Dong'. That is to say, it is only the most recent article, it may be at the top, and it is a subject that is not so much topic, but it has been exposed at the top. This is a phenomenon that occurs because an existing search does not consider only the fact that the search keyword is included and does not consider the contents of the search result at all.

According to the present invention, there is provided a method of searching keywords using big data through an electronic terminal, comprising: setting a big data database to be searched in the electronic terminal; Receiving a search keyword through an input unit of the electronic terminal; Wherein the electronic terminal comprises: a step of selecting data including the search keyword from among the data included in the big data database; Counting the number of words or phrases included in the selected data; The electronic terminal ranking the counted words or phrases in order of appearance frequency; The electronic terminal selecting a plurality of ranked words or phrases as a plurality of related search words; And visually displaying the selected plurality of related search terms on a display of the electronic terminal.

Preferably, the displaying step displays a higher-level related keyword in a larger size.

Advantageously, said step of displaying is displayed in a larger circle or larger polygon box with higher associated query terms.

Preferably, in the displaying step, the color of the text is changed according to the ranking, or the text of some ranking is horizontally written and the text of some ranking is arranged vertically.

Preferably, the big data database is an SNS article.

Preferably, a specific company or business is evaluated by the keyword search.

According to the present invention, there is provided an electronic device for searching keywords using big data, comprising: a setting unit for setting a big data database to be searched; An input unit for receiving a search keyword; A control unit; And a display, wherein the control unit selects data including the search keyword among the respective data included in the big data database, counts the number of words or phrases included in the selected data, Counting the words or phrases counted in ascending order of frequency, selecting a plurality of words or phrases ranked as the plurality of related words, and causing the display to visually display the plurality of selected related words A keyword search electronic device is provided.

Preferably, the display control of the control unit causes the display unit to display a higher-related related word in a larger font.

Advantageously, said display control of said control unit causes said higher associated query to appear in a larger circle or larger polygon box.

Preferably, the display control of the control unit causes the color of the letters to be different according to the rank, or to horizontally write the letters of some rankings and vertically to write letters of some rankings.

Preferably, the big data database is an SNS article.

Preferably, a specific company or business is evaluated by the keyword search.

According to the search of the present invention, since the related search word is the most commonly used word with the search keyword, it provides a very accurate result for the search keyword.

According to an aspect of the present invention, the first associated query is displayed in the largest size, the second related search term is displayed in the next larger size, And displaying the search term in the next larger size (i.e., smaller than the second associated search term). Other visual arrangements are also acceptable.

Through such a big data search, it is possible to easily understand what reference is made to the search keyword on the Internet (more narrowly on the SNS) and what matters about the search keyword And it is possible to solve the conventional problem that the garbage or noise can be retrieved at the top (that is, the problem of acquiring the erroneous information) without additional effort of the person as in the conventional search.

1 is a diagram showing an example of a search result according to the present invention.
2 shows an example of a flow chart of a search according to the present invention.
FIG. 3 shows the flow of the search according to the present invention in four stages.
4A to 4C illustrate a process of inputting a 'user request association word' mentioned in FIG. 3 as an example of making a search according to the present invention in the form of a smartphone application.
5A to 5C show an example of the 'real-time data collection' process mentioned in FIG. 3 as an example of a search according to the present invention in the form of a smartphone application.
6A and 6B illustrate a process of searching for data according to the present invention in the form of a smartphone application.
FIG. 7 shows an example of making a search according to the present invention in the form of a smartphone application, and the 'visualization' process mentioned in FIG.
8 is a diagram for collecting portal issue word associations.
Figure 9 shows an example of a device according to the invention.

Hereinafter, a system and a method according to the present invention will be described with reference to the drawings.

1 is a diagram showing an example of a search result according to the present invention.

Existing Internet keyword searches, for example, searches data stored in a related server of a portal site or data on the Internet, and finds and presents data containing (or related to) the keyword. Most of the materials presented are web pages, picture files, video files, and so on. In many cases, except for pictures, videos, etc., homepage links with text data are often searched.

The present invention uses big data as a database and displays related keywords by differentiating the sizes of characters so as to intuitively recognize the search results, rather than simply displaying them in the form of letters.

For example, when the name of a specific candidate is entered in an existing search site (Naver, Daum, Google, etc.) in connection with the election of a parliamentarian, a homepage link material such as a newspaper article about the person will be listed. However, according to the present invention, when a specific candidate name is input, the search database is different from conventional ones as big data. And, it shows not the web page is searched, but the one that is mentioned most often based on the number mentioned in big data such as SNS. Show Top can also be shown at the top of the search results, but more preferably, it shows the most commonly mentioned in the largest text.

If you search for a specific candidate name (Hong Gil Dong) according to the present invention, various keywords may appear. For example, Candidate Hong Gil Dong is widely exposed to the media in relation to military service corruption. , The method and apparatus of the present invention for the search keyword "Hong Kil-Dong", as a related keyword (i.e., a search result), is referred to as the first related keyword (i.e., 'Hong Gil-dong' and the keyword most commonly referred to in the SNS article). For example, if some SNS articles (Twitter mentions, etc.) are said to have "Hong Gil-dong has been suspected of military service corruption," the related keyword of "Hong Gil-dong" is limited to this SNS article, The number of matches (number of matches) of 'military service irregularities' with respect to the search keyword 'Hong Gil Dong' as the number of articles such as this article (eg, "Candidate Hong Gil is under investigation by military service corruption" It will be getting higher. If there are many SNS articles using similar vocabulary, or if the SNS is much enlarged / reproduced (for example, the Twitter function of Twitter, etc.), the degree of association (matching frequency) will gradually increase.

Next, if a keyword frequently referred to together with the keyword 'Hong Gil-Dong' is a 'commitment', a 'commitment' may be selected as a second related keyword. In this way, the keyword 'Hong Gil - dong' is ranked along with the keywords related to the big data such as SNS articles. In this case, since the search object is not limited to the article stored in the specific server such as Naver's portal site but is big data, it can be expected that the accuracy of such search will be considerably high. The criterion for selecting a related keyword may be, for example, a keyword limited to only a keyword mentioned over a thousand times with a search keyword, and may be ranked among the keywords.

Through the above description, it can be seen that this is very different from the search in the existing portal site. The search on the existing portal site may be performed only on the data collected in advance by the portal site (for example, Naver, next), or may be search the Internet on the whole even if it is not collected in advance (for example, Google). Although the present invention can be applied to general Internet search, it is preferable that only big data is targeted. Big data refers to SNS articles such as Twitter and Facebook as an example. Of course, in some cases, the search may be performed including data (for example, internet news articles, blog articles, etc.) on the Internet as described above. However, the difference from the existing search is that not searching for specific data related to a search keyword (i.e., a specific press article, a specific blog article, a specific image, a specific video, etc.) The frequency of other related keywords appearing with the keyword (that is, within the same paragraph or the same article) is found, and the ranking is determined and the ranking is shown. In the past, the top most search result (homepage link) may or may not represent 'Hong Gil Dong'. That is to say, it is only the most recent article, it may be at the top, and it is a subject that is not so much topic, but it has been exposed at the top. This is a phenomenon that occurs because an existing search does not consider only the fact that the search keyword is included and does not consider the contents of the search result at all. However, according to the search of the present invention, since the first associated search word is the most commonly used vocabulary with the search keyword, it provides a very accurate result for the search keyword.

According to an aspect of the present invention, the first associated query is displayed in the largest size, the second related search term is displayed in the next larger size, And displaying the search term in the next larger size (i.e., smaller than the second associated search term). In this case, the first associated keyword does not necessarily have to be at the uppermost position, but may be displayed at the center of the search terminal screen or may be displayed at another position. It is easy to know that the first related keyword is displayed anywhere on the screen because the font size is the largest.

In this manner, the font size and the degree of associativity may be displayed, the degree of association may be displayed in the size of the circle, the first associated query may be inserted in the largest circle, and the second related search term may be inserted in the next larger circle. Needless to say, the criteria of circle size and letter size may be used in combination. The largest position of the circle can be displayed at the uppermost position and gradually become smaller toward the bottom, or the largest one may be displayed at the leftmost position and gradually become smaller toward the rightmost end, or displayed randomly on the screen, It can be judged. The text does not necessarily have to be horizontally written, but any associated query may be horizontally, and any associated query may be vertical, so that it can be visually and intuitively grasped. Also, the colors may be different for each related search word.

Through such a big data search, it is possible to easily understand what reference is made to the search keyword on the Internet (more narrowly on the SNS) and what matters about the search keyword have.

Alternatively, a service may be provided that not only lists the plurality of related search terms, but also displays, for example, in which phrases the first related search term is displayed when the first related search term is clicked on the screen of the search terminal.

That is, if the first associated keyword is "military corruption" and the second associated keyword is "pledge" with "Hong Gil Dong" as a search keyword, the word "military corruption" The big letters, and the rest of the related search terms, will be listed in the font size for each rank. Each associated search term can be configured to be clickable, and by clicking on a particular associated search term (e.g., "Commitment"), you can see in detail which words " . The specific text is a piece of big data, and it may be a normal SNS text, but it may be searched to include articles of a normal Internet site (for example, Internet media, blog, etc.) as needed.

An example of the case of searching for a celebrity "Tuya" is shown in FIG. 1, although the example of a member of the National Assembly was taken as an example. The operation principle is the same as described above.

That is, the most related words (first associated search word) for the search term "Tuya" are "Kim Ji-hye" and are indicated by large letters in the center of FIG. The next most related word (second associated query) is "Sugar Man" and is shown in the middle of FIG. 1 as a large letter at the bottom. The next most related word (the third associated query) is "nystagmus ", which is slightly overlaid in the middle of FIG. Other related search terms (that is, below the fourth related search term) can be seen in the form of a plurality of characters enclosing the letter "Ji-hye" in the middle.

In service of the method or apparatus according to the present invention, the general service version may provide only an associated search term, and the paid service version may be configured to view a specific article (which phrase is derived) by clicking an associated search term.

The Big Data may be provided through a linkage / agreement with a company such as Teradata Corporation, for example. Through this linkage, individuals will be able to see the big data of public services.

As another example, it can be assumed that, for example, a specific restaurant name is searched. According to the present invention, when searching for a specific restaurant name, for example, an associated search word "weave" may appear and an associated search term "delicious" Both of them come out a lot, but if they are the most related to "weave", this is the dominant one. That is, when a specific restaurant name is searched, if the first related search word is "weave" and the second related search word is "delicious", then the user of the present invention is informed that the particular restaurant is delicious, Can be obtained. According to the existing search, you can not read the review of the restaurant that is on the top. This is a time-consuming task, and the review is not necessarily fair. That is, in a conventional search in which a publicity blog entry may be retrieved at the top, there is a high possibility that the blog information will have erroneous information about the specific restaurant. However, according to the search of the present invention, for example, since the name of the restaurant and the word "weave" simultaneously appear in a large number of articles (data), a completely different effect from the conventional search in which a single article is read It is self-evident.

Particularly, in the case of using a graphic in the present invention, various visual and intuitive interfaces are provided through, for example, displaying the size of the association with different sizes of circles, horizontal writing, vertical writing, and color mixing, Easy and accurate information can be obtained.

According to the present invention, as a result, it is possible for an individual to evaluate restaurants, businesses, and the like.

In the conventional search, since the contents must be read by reading articles and the like with the result of the keyword search through the human eyes, a great deal of human power is required in the end. However, according to the present invention, since the machine reads the most frequently read letters, it does not need much human power. If a lot of searches are made, a large text is displayed. Therefore, it is very easy to grasp the degree of association and discrimination power is given. Not only large letters but also highlights, color difference, or any other visual effect may be given or combined.

The object of the big data according to the present invention can be variously diversified such as an article, a blog, an SNS, and a Twitter. Some or all of them can be used and it is possible to apply them to a new database system as needed.

The analysis for the search (related keyword selection) of the present invention can use a technique such as machine learning (deep learning) or the like as a big data analysis technique, and the method and apparatus of the present invention are web- Or an application based on a smart phone.

On the other hand, it is also possible to re-search part of the search result (that is, showing the related search word) again, or link it to Twitter or blog. When searching for "Hong Kil-dong" and searching for "military service corruption" is the first related search word, clicking on "military service corruption", now the "military service corruption" becomes search keyword and searching for the related search word.

The term 'related search word' itself may be a somewhat broad term, but it should not be understood simply by the meaning of the word, and it is not necessary to understand the precondition of the present invention as described above, It is not a term alone to judge whether or not an object or method is different because it refers to an associated search term in which the conditions are satisfied.

That is, according to the conventional technology, it is impossible or difficult for an individual to evaluate a company or a mart or the like, but according to the present invention, each article constituting the big data becomes meaningful evaluation data, It is easy to see that it is a different way of searching for and displaying the entries. For example, in the past, if a new credit card is issued, the service is only searched and evaluated internally. However, according to the present invention, it is obvious that the service can be utilized in a totally different manner as a service allowing the general public to view the data.

According to the present invention, as an application, a quiz game may be performed to predict what the first associated keyword will appear when a search term "Hong Gil Dong" is inserted. Or quizzes that predict when a search query will put "corruption" when you put in a query. It will also serve as a simple and fun play among young people.

In the conventional search, although the amount of data is large, there is a difficulty in access and processing in fact. It is also difficult to do on a PC, not a smartphone. For example, in order to avoid such a hassle (that is, useless information is searched), there is a person who enters and searches the homepage of a newspaper company. The present invention may be a good tool for people who want to filter the noise.

At present, in the case of newspaper articles on the web, search results are arranged in order of importance. However, as you go from the first article to the last article, it seems that similar words have been used more and more, and in the end, it is hard to know the subtleties until each individual article is viewed. However, according to the present invention, since the words of each article (when the big data is the newspaper article database) are counted, it has a new meaning different from the conventional search. That is, the center keyword can be selected according to the occurrence frequency of words. The central keyword may be regarded as the same concept as the above-described related search word.

On the other hand, although the concrete form of the visualization has been described above, the present invention is not limited to this, and it is possible to give discrimination power in various visual forms.

The word count (i.e., association check) process can use various algorithms and a weighting scheme can be used. And, if it is judged that some data in big data is commercial (data), it is also possible to exclude it from word count. The judgment of commerciality may be made, for example, by checking whether a predetermined (predefined) advertisement phrase is included in the phrase of the data, or by checking whether a specific homepage address is included, and so on .

2 shows an example of a flow chart of a search according to the present invention.

The present invention provides a method and apparatus for searching keywords using big data through an electronic terminal.

For example, a big data database to be searched in the electronic terminal is set (step 201). Then, the search keyword is input through the input unit of the electronic terminal (step 202).

The electronic terminal selects data including the search keyword among the respective data included in the big data database (step 203). The electronic terminal counts the number of words or phrases included in the selected data (step 204). Then, the electronic terminal places the counted words or phrases in order of appearance frequency (step 205).

The electronic terminal selects a plurality of ranked words or phrases as a plurality of associated search terms (step 206). The selected plurality of related search terms are visually displayed on the display of the electronic terminal (step 207).

The present invention may operate in a PC or the like connected with a big data database via a network, or may operate in a device such as a smart phone.

FIG. 3 shows the flow of the search according to the present invention in four stages.

The present invention can be seen as a process of 'inputting a user request related word', 'real time data collection', 'data analysis', and 'visualization'.

Each of these will be described below.

4A to 4C illustrate a process of inputting a 'user request association word' mentioned in FIG. 3 as an example of making a search according to the present invention in the form of a smartphone application.

A search window (input window) is displayed at the top of FIG. 4A, and twenty real-time search words are displayed in the bottom of the portal site (for example, Naver, Next, etc.).

For example, the main page of an app according to the present invention is written in HTML format for web application. In the HTML document, a user's action is configured as a java script to pass the value to the system internal program.

The user can input a word desired to be searched through the input window displayed at the upper part of FIG. 4A, and the inputted contents can be transmitted to the internal collection program by clicking the enter key or the analysis button below.

Referring to the lower part of FIG. 4A, the upper 20 list windows of the real-time search terms of portal sites are arranged at the lower end separately from the user input part of the upper part of the page. It automatically collects the real-time query ranking, collects and analyzes the web data related to the word, and places it on the main screen accessed by the user, thereby giving additional value to the convenience of the user.

The user request association function (i.e., the function related to the input window at the upper part of FIG. 3) provides a function of allowing a user to access a main page of an app according to the present invention to input at least one to a plurality of desired words . Technically, it is possible to receive text as an input value, and a plurality of words can be separated into a blank space and transmitted as an input value. If you press the Analyze button without entering a word or click the Enter button, you will be prompted to enter a word. This can be implemented, for example, via the code shown in FIG. 4B.

Next, when the user enters a normal word and clicks the analysis button or the enter key, the internal search.jsp file is called to transmit the inputted parameter value. search.jsp passes the value of the argument passed back to the data collection program. This can be implemented, for example, through the code shown in FIG. 4C.

FIGS. 5A to 5C illustrate a process of 'real-time data collection' mentioned in FIG. 3 as an example of making a search according to the present invention in the form of a smartphone application.

In order to collect real-time data on the Internet, a web scraping engine is executed to search documents containing a user-entered word and store the result in the form of html or plain text. The result of the search is to use the word entered by the user as the file name. This can be implemented, for example, through the code shown in FIG. 5A.

The initial data collected through the web scraping engine contains html tags and unnecessary characters that can not be immediately analyzed, so data filtering is performed to refine these unnecessary contents. The file that has been filtered is in the form of a plain text file. This can be implemented, for example, through the code shown in FIG. 5B.

5C, the present invention collects data from various sources such as a web, an SNS, a blog, and the like to find a related word of a user request (that is, a word input in the input window of FIG. 4A) It is possible to collect related words of issue words (i.e., words 1 to 10 and words 11 to 20 at the bottom of FIG. 4A).

6A and 6B illustrate a process of analyzing data according to an exemplary embodiment of the present invention. Referring to FIG.

Data analysis is done using R 's KoNLP package, which is open source software. It performs a task of receiving a filtered text file and deleting unnecessary symbols that can not be removed by the filtering program. This can be implemented, for example, via the code shown in FIG. 6A.

Next, rewrite the extracted nouns as a list and measure the frequency of each noun. Outputs the top 100 words as a separate csv file for secondary analysis of the frequency of nouns and nouns. 20 of the top 100 words are extracted and used as keywords in the results. This can be implemented, for example, via the code shown in FIG. 6B.

FIG. 7 shows an example of making a search according to the present invention in the form of a smartphone application, and the 'visualization' process mentioned in FIG.

Visualization uses R's word cloud package to output to the png file format. In the visualization of big data analysis, word cloud places the word with the highest frequency in the center at the center, and then outputs the word with the next frequency smaller. The output result is closer to the circle as the number of words increases. This can be implemented, for example, through the code shown in FIG. 7A.

The result of completing the visualization work is transferred to the first requested user screen. The results delivered include the resulting png file and a list of the top ten words. The search.jsp file creates the html form of the received png file name and word list and delivers it to the user's screen. The top ten words of the analysis result are linked to the search window of the portal site (for example, Naver) for the user's re-searching convenience.

By performing such a process, for example, if "Two Night" is input in the input window of Fig. 4A, the search result as shown in Fig. 1 can be obtained.

4A, the list of a predetermined number of issue words (for example, 20) to be displayed on the main screen may be displayed every five minutes, for example, Automatically accesses the portal and collects word lists. In order to maintain the real-time nature of the information provided, information is collected every two hours in the case of the same word. The collected list is delivered to the web scraping program and the analysis result is stored in the server and provided to the user. This can be implemented, for example, through the code shown in Fig.

Figure 9 shows an example of a device according to the invention.

And provides an electronic device for searching keywords by using big data. The apparatus includes a setting unit 301 for setting a big data database to be searched, an input unit 303 for inputting a search keyword, a control unit 305, and a display 307.

The control unit 305 selects data including the search keyword from among the respective data included in the big data database, counts the number of words or phrases included in the selected data, and outputs the counted word or phrase as the appearance frequency A plurality of ranked words or phrases are selected as a plurality of related search words, and the display 307 can control to visually display a plurality of selected related search words.

The control unit 305 may be, for example, a CPU of a PC or a CPU of a smart phone.

Although the present invention has been described with reference to specific embodiments, many modifications may be made by those skilled in the art without departing from the spirit and scope of the present invention as defined by the appended claims. Of course. It is to be understood that various other modifications may be made without departing from the scope of the present invention.

Claims (12)

A method for searching keywords using big data through an electronic terminal,
Setting a big data database to be searched in the electronic terminal;
Receiving a search keyword through an input unit of the electronic terminal;
Wherein the electronic terminal comprises: a step of selecting data including the search keyword from among the data included in the big data database;
Counting the number of words or phrases included in the selected data;
The electronic terminal ranking the counted words or phrases in order of appearance frequency;
The electronic terminal selecting a plurality of ranked words or phrases as a plurality of related search words; And
Visually displaying the selected plurality of related search terms on a display of the electronic terminal
A keyword search method. The method according to claim 1,
Wherein the step of displaying the keyword includes displaying the keyword in a larger size as the related keyword having a higher ranking is displayed. The method according to claim 1,
Wherein the displaying step displays in the larger circle or larger polygon box the higher the associated ranked query is. The method according to claim 1,
The displaying step may include a keyword searching method in which the color of the character is changed according to the ranking, or the character of some ranking is written horizontally and the letters of some ranking are arranged vertically, The method according to claim 1,
Wherein the big data database is an SNS font. 6. The method according to any one of claims 1 to 5,
And evaluating a specific company or business by the keyword search. An electronic device for searching a keyword using big data,
A setting unit for setting a big data database to be searched;
An input unit for receiving a search keyword;
A control unit; And
display
/ RTI >
The control unit selects data including the search keyword from among the respective data included in the big data database, counts the number of words or phrases included in the selected data, and stores the counted word or phrase A plurality of words or phrases ranked in the order of appearance frequency, a plurality of words or phrases ranked as the plurality of related words, and controlling the display to visually display the plurality of selected related words. . 8. The method of claim 7,
Wherein the display control of the control unit causes the display unit to display, in a large font, a higher-ranked associated keyword. 8. The method of claim 7,
Wherein the display control of the control unit causes the display unit to display in a larger circle or larger polygon box as the higher ranked associated query is displayed. 8. The method of claim 7,
Wherein the display control of the control unit causes the color of the text to be different according to the ranking, or to horizontally place the text of some ranking and to vertically position the text of some ranking. 8. The method of claim 7,
Wherein the big data database is an SNS article. 12. The method according to any one of claims 7 to 11,
And to evaluate the specific company or business by the keyword search.

Images