WO2010036012A2 - 인터넷을 이용한 의견 검색 시스템, 의견 검색 및 광고 서비스 시스템과 그 방법 - Google Patents
인터넷을 이용한 의견 검색 시스템, 의견 검색 및 광고 서비스 시스템과 그 방법 Download PDFInfo
- Publication number
- WO2010036012A2 WO2010036012A2 PCT/KR2009/005405 KR2009005405W WO2010036012A2 WO 2010036012 A2 WO2010036012 A2 WO 2010036012A2 KR 2009005405 W KR2009005405 W KR 2009005405W WO 2010036012 A2 WO2010036012 A2 WO 2010036012A2
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- opinion
- search
- information
- negative
- positive
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0255—Targeted advertisements based on user history
- G06Q30/0256—User search
Definitions
- the present invention relates to an opinion retrieval system, an opinion retrieval and advertisement service system using the Internet, and a method thereof, and more particularly, to automatically extract and analyze user opinion information scattered on various websites existing on the Internet. It provides opinion search service to check search and statistics by negative opinions, and provides customized advertisement service suitable for each opinion search user with user opinion information scattered on various websites existing on the Internet. They can easily search and monitor other users 'opinions on specific keywords at a glance, and users can easily search and monitor other users' opinions on specific keywords at a glance. Sleeping in the field Get more effective advertising effect for the goods, and further comments by the goods in the Internet can be further improved purchase probability of the search relates to a system, feedback and search advertising system and method.
- an object of the present invention is to automatically extract and analyze user opinion information scattered on various websites existing on the Internet to confirm search and statistics by positive / negative opinions
- an object of the present invention is to automatically extract and analyze user opinion information scattered on various websites existing on the Internet to confirm search and statistics by positive / negative opinions
- Another object of the present invention is to provide a customized advertisement service suitable for each opinion search user simultaneously with user opinion information scattered on various websites existing on the Internet. Not only can they easily search and monitor their opinions at a glance, but also advertisers can get more efficient advertising effects on their products and further improve the probability of purchasing them. It provides an opinion retrieval and advertisement service system and its method.
- a first aspect of the present invention includes a first server for collecting web document data existing on the Internet; A language processing module that separates the collected web document data into sentence units and performs linguistic processing on each of the separated sentences to extract linguistic features; An opinion / non-computation classification module for classifying opinion / non-comment sentences using the linguistic qualities of the extracted sentences; An opinion expression division module for dividing the linguistic qualities of the divided opinion sentences into positive / negative opinion expressions; A second server indexing the opinion information of the corresponding web document to be stored according to linguistic features of the divided opinion sentences; And a web server that receives a specific keyword transmitted from a user terminal connected through the Internet, searches for opinion information of a web document related to the specific keyword in association with the second server, and displays the opinion search result on the screen of the corresponding user terminal. It is to provide a opinion retrieval system using the Internet.
- the data storage module may further include a data storage module configured to extract and store at least one information data of text, image, or video information required from the web document data collected through the first server.
- the language processing module separates the general document data including the opinion / non-comment sentences set in advance with the collected web document data in sentence units, and performs linguistic processing on each of the separated sentences. You can extract the qualities.
- the opinion indexing information storage module further stores the summary information of the corresponding opinion sentences for each of the opinion sentences indexed through the second server and the basic and opinion information of the corresponding web document into a database and is stored in a database. May be included.
- the basic and opinion information of the web document may include a title, a text, an analyzed text, a creation date, a tag, a URL, an image, a video, the number of positive / negative expressions, the overall positive / negative degree, and each positive / negative expression.
- Location information of the beginning and end of the information, object keyword information that can be the target of the opinion vocabulary, object keyword and opinion expression relationship information, or type information to which each object keyword belongs may be composed of at least one information.
- the language processing may be performed by Morpheme Analyze or Segmentation.
- the web server may be displayed on the screen of the user terminal to selectively check the overall opinions, positive / negative opinions related to the particular keyword.
- the web server determines the ratio of positive / negative opinion expression in the overall opinion search result related to the specific keyword, or the ratio of positive / negative opinion expression in each opinion information related to the specific keyword. Can be displayed on the screen.
- the web server may list the opinion search results related to the specific keyword in order of importance or time and display them on the screen of the corresponding user terminal.
- the importance level is determined based on the degree of relevance and opinion expressed by the specific keyword in the corresponding web document, and is limited to the entire time range or the specific time range, and the time order is in the order in which the corresponding web document is generated. As a result, it may be determined in ascending / descending order, and may be limited to the entire time range or a specific time range.
- the web server may display a comment input window on the screen of the corresponding user terminal to add the opinion of the corresponding opinion search user in the form of a comment on the opinion contents of the web document related to the specific keyword.
- the web server may display the opinion search result related to the specific keyword on the screen of the corresponding user terminal by emphasizing the part expressed as positive / negative with the specific keyword in a specific expression.
- the web server may analyze the positive / negative opinion part according to the user's selection of the opinion search result articles related to the specific keyword, and highlight it in a specific expression and display it on the screen of the corresponding user terminal.
- the specific expression may be made of underline, bold text, or at least one emphasizing expression of various colors.
- the web server may display a change in the positive / negative ratio in the form of a graph on the screen of the corresponding user terminal in accordance with the degree of positive / negative opinion expression of the opinion search result related to the specific keyword.
- the web server may display a positive / negative ratio on the screen of the corresponding user terminal for each opinion search result related to the specific keyword.
- the web server may be displayed on the screen of the user terminal to select the user's consent / objection to the opinion search results associated with the particular keyword.
- the web server may monitor in real time the generation of affirmative / negative opinion related to a specific keyword registered by the user and notify the corresponding user terminal.
- a second aspect of the invention includes the steps of: (a) collecting web document data residing on the internet; (b) separating the collected web document data into sentence units and performing linguistic processing on each of the separated sentences to extract linguistic features; (c) dividing the opinion / non-comment sentences using the linguistic qualities of the extracted sentences; (d) dividing the linguistic qualities of the divided opinion sentences into positive / negative opinion expressions; (e) indexing the opinion information of the corresponding web document to be stored according to linguistic qualities of the divided opinion sentences; And (f) searching for opinion information of a web document related to a specific keyword transmitted from a user terminal connected through the Internet, and displaying the opinion search result on a screen of the corresponding user terminal. It is.
- the general document data including the opinion / non-comment sentences set in advance together with the collected web document data are separated in sentence units, and language processing is performed on each of the separated sentences. You can extract linguistic qualities.
- step (e) the summary information of the corresponding opinion sentence for each of the linguistic qualities of the indexed opinion sentences and the basic and opinion information of the corresponding web document are made into a database and stored in a separate storage module. It may further comprise a step.
- the language processing may perform morphological analysis or spacing.
- the overall opinion related to the specific keyword and positive / negative opinion contents may be selectively displayed. Can be.
- the ratio of positive / negative opinion expression in the overall opinion search result related to the specific keyword, or The ratio of positive / negative opinion expression in each opinion information related to the specific keyword may be displayed.
- step (f) when displaying the opinion search results associated with the particular keyword on the screen of the user terminal, it may be displayed in the order of importance or time.
- the importance level is determined based on the degree of relevance and opinion expressed by the specific keyword in the corresponding web document, and is limited and applied to the entire time range or the specific time range, and the time sequence is in the order in which the corresponding web document is generated. Therefore, it can be determined in ascending / descending order and can be applied to the entire time range or to a specific time range.
- the opinion search user of the opinion search user in the form of a comment on the opinion contents of the web document related to the specific keyword. You can display a comment box to add a comment.
- the part expressed as positive / negative together with the specific keyword can be highlighted and displayed in a specific expression. have.
- the particular expression may be underlined, bold or at least one of the various colors.
- the opinion search result related to the specific keyword is displayed on the screen of the corresponding user terminal, positive / negative opinion according to the user's selection for the opinion search result articles related to the specific keyword.
- it can be displayed by highlighting at least one of underline, bold text, or various colors.
- the change of the positive / negative ratio of each time according to the degree of positive / negative opinion expression is displayed in graph form. I can do it.
- the positive / negative ratio may be displayed for each detailed item of the specific keyword.
- the method may further include the step of monitoring in real time the generation of affirmative / negative opinion related to a specific keyword registered by the user and notifying the corresponding user terminal.
- a third aspect of the present invention is to provide a recording medium on which a program for executing the above-described opinion retrieval method using the Internet is recorded.
- opinion information DB that stores the opinion information of the corresponding web document for each language feature of the opinion sentence; An advertisement information DB in which advertisement information for each keyword is stored; And receiving a specific keyword transmitted from a user terminal connected through the Internet, interworking with the opinion information DB and the advertisement information DB, searching for opinions and advertisement information of a web document related to the specific keyword, and providing opinions on the screen of the corresponding user terminal. It is to provide an opinion retrieval and advertisement service system using the Internet that includes a web server displaying advertisement information related to the search result articles.
- the opinion information DB is preferably stored in a database (DB) of the summary information of the corresponding opinion sentence for each language and the basic and opinion information of the web document.
- DB database
- the basic and opinion information of the web document may include a title, a text, an analyzed text, a creation date, a tag, a URL, an image, a video, the number of positive / negative expressions, the overall positive / negative degree, and each positive / negative expression.
- Location information of the beginning and end of the information, object keyword information that can be the target of the opinion vocabulary, object keyword and opinion expression relationship information, or type information to which each object keyword belongs may be composed of at least one information.
- the opinion information stored in the opinion information DB is separated into sentence units for the web document data existing on the Internet, and language processing is performed on each of the separated sentences to extract linguistic features. Distinguish the opinion / non-computation sentences using the linguistic qualities of each sentence, and classify the linguistic qualities of the divided opinion sentences into positive / negative expressions, and by linguistic qualities of the divided opinion sentences.
- the opinion information of the web document may be indexed and stored.
- the language processing may be morphological analysis or spacing.
- the advertisement information is searched by the advertiser, and as a result, at least one of the advertisement link, advertisement phrase, or advertisement image information for each keyword for each keyword or opinion search type is converted into a database (DB). Can be stored.
- DB database
- the opinion search types may be made of any one type selected from among whole opinion contents, positive / negative opinion contents, or positive / negative opinion portion analysis contents of opinion search result articles.
- the web server is displayed on the screen of the user terminal to selectively check the overall opinions, positive / negative opinions related to the specific keyword, positive / negative in the overall opinion search results associated with the specific keyword
- the advertisement information related to the ratio of opinion expression or the ratio of positive / negative opinion expression in each opinion information related to the specific keyword may be displayed on the screen of the corresponding user terminal.
- the web server displays the advertisement information related to the positive opinion content related to the specific keyword on the screen of the corresponding user terminal, or the search user for the negative opinion content of the web document related to the specific keyword.
- the input window can be displayed on the screen of the corresponding user terminal so that the comment text can be posted.
- the web server analyzes the positive / negative part of the opinion search result articles related to the specific keyword according to the user's selection, and displays the advertisement information related to the analyzed opinion part on the screen of the corresponding user terminal.
- the web server may provide a part of the advertising revenue to the content provider providing each opinion search result article according to the search ranking of the corresponding content, whether the search user is selected, and the number of recommendations for the corresponding content.
- the fifth aspect of the present invention includes the steps of: (a) storing opinion information of a corresponding web document in a separate opinion information DB for each language feature of the opinion sentence; (b) storing the advertisement information for each keyword in a separate advertisement information DB; And (c) retrieving opinions and advertisement information of a web document related to a specific keyword transmitted from a user terminal connected through the Internet in the opinion information DB and advertisement information DB, respectively, together with the opinion search result articles on the screen of the corresponding user terminal.
- the present invention provides a method of searching for opinions and advertising using the Internet, including displaying related advertisement information.
- a database (DB) may be stored and stored in the opinion information DB as a database (DB) of the summary information of the corresponding opinion sentence for each linguistic feature and the basic and opinion information of the corresponding web document.
- the opinion information stored in the opinion information DB is separated into sentence units for web document data existing on the Internet, and the linguistic processing is performed on each of the separated sentences.
- the linguistic processing is performed on each of the separated sentences.
- step (b) at least any one of a search preset by the advertiser in the advertisement information DB and the result of the keyword-specific ad link, advertisement phrase or advertisement image information for each keyword or opinion search type.
- Advertising information can be stored in a database (DB).
- the opinion search types may be made of any one type selected from among whole opinion contents, positive / negative opinion contents, or positive / negative opinion portion analysis contents of opinion search result articles.
- the advertisement information related to the opinion search result articles related to the specific keyword is displayed on the screen of the corresponding user terminal
- the overall opinion and positive / negative opinion contents related to the specific keyword are selectively selected. Display on the screen of the corresponding user terminal so as to be identified, and express the negative / negative opinion in all opinion search results related to the specific keyword, or express the positive / negative opinion in each opinion information related to the specific keyword.
- the advertisement information associated with the ratio may be displayed on the screen of the corresponding user terminal.
- the advertisement information related to the opinion search result articles related to the specific keyword is displayed on the screen of the corresponding user terminal
- the advertisement information related to the positive opinion contents related to the specific keyword is displayed.
- the input window may be displayed on the screen of the corresponding user terminal so as to be displayed on the screen of the corresponding user terminal or to post a comment of the search user with respect to the negative opinions of the web document related to the specific keyword.
- the user selects the opinion search result articles related to the specific keyword.
- the positive / negative opinion portion may be analyzed, and advertisement information related to the analyzed opinion portion may be displayed on the screen of the corresponding user terminal.
- step (c) providing a part of the advertising revenue to the content provider providing each opinion search result article according to the search ranking of the content, whether the search user is selected, and the number of recommendations for the content. It may further include.
- a sixth aspect of the present invention is to provide a recording medium on which a program for executing the above-mentioned opinion retrieval and advertisement service method using the Internet is recorded.
- the user opinion information scattered on various websites existing on the Internet are automatically extracted and analyzed to search and statistics by positive / negative opinions.
- users can easily search and monitor other users 'opinions on a particular keyword at a glance, and greatly reduce the time spent in searching for other users' opinions. There is an advantage.
- marketers, stock investors, corporate value evaluators, etc. of each company can check the opinions of various users on the company or goods existing on the vast Internet at a glance, and know the opinions of the users. In order to reduce the cost of surveys and consulting firms, it is possible to effectively extract the opinions and statistics of each user.
- the opinion search users different from a specific keyword Not only can users easily search and monitor their opinions at a glance, but also advertisers can get more efficient advertising effect on their products, and moreover, improve the probability of purchasing the products. have.
- FIG. 1 is a block diagram illustrating an overall opinion retrieval system using the Internet according to an embodiment of the present invention.
- FIG. 2 is an overall flowchart illustrating a method for searching for opinions using the Internet according to an embodiment of the present invention.
- FIG. 3 to 6 are screen configuration diagrams for explaining the opinion search results applied to an embodiment of the present invention
- FIG. 5 is a comment search result related to a specific opinion search keyword (nomnomnom)
- FIG. 6 is a screen configuration diagram showing the function of a page analyzed for opinions
- FIG. 6 is a screen configuration provided to select a user's consent / opposition for opinion search result articles related to a specific opinion search keyword (nomnomnom). It is also.
- FIG. 7 is a block diagram illustrating an overall opinion search and advertisement service system using the Internet according to another embodiment of the present invention.
- FIG. 8 is a flowchart illustrating a method of searching for opinions and advertising services using the Internet according to another embodiment of the present invention.
- 9 to 12 are screen configuration diagrams for explaining the results of opinion search and advertisement service applied to another embodiment of the present invention.
- FIG. 1 is a block diagram illustrating an overall opinion retrieval system using the Internet according to an embodiment of the present invention.
- the opinion retrieval system using the Internet data collection server 100, language processing module 200, opinion / non-comment classification module 300, opinion expression classification Module 400, indexing server 500, opinion indexing information storage module 600, opinion search module 700, web server 800, and user terminal 900.
- the data collection server 100 performs a function of collecting various web document data existing on the Internet 10. That is, the data collection server 100 receives in real time download (Hyper Text Markup Language) information of each Web site (Web Site) existing on the Internet (10).
- Web Site Web Site
- the data collection server 100 is at least any one of the information required in the web document data downloaded as described above, such as information (Text), image (Image), video (Video), etc.
- the data may be extracted and stored in a separate data storage module 150.
- the data collection server 100 may collect and collect web document data including opinion information data (ie, general sentence / document data and information data given affirmative / negative evaluation thereof) as shown in Table 1 below. It may be.
- opinion information data ie, general sentence / document data and information data given affirmative / negative evaluation thereof
- the specific web document data including the opinion information data is selected and machine learning algorithms (for example, SVM and K-) described later are selected.
- NN, Bayseian, etc.) to generate a web document screening model, and then use the generated web document screening model to selectively collect only web document data including opinion information data from the entire Internet web page. do.
- Table 1 expression score Opinion Content ⁇ 10 It is interesting and sends it ⁇ 10 Report stories of 'smart' people ⁇ 8 The daily plucking of wise people! Declaration ⁇ 9 Soaked in my uncle's charm ... Shingo ⁇ 8 Report stories of ordinary people, not smart people ⁇ 10 It is a love story with good acting and good content. Uncle is so attractive ⁇ ???? Declaration ⁇ 10 It was a very spectacular story. Declaration ⁇ 10 I watched with no expectation, it was a heart warming movie all the time. It's fun. Report it. ⁇ 6 It's warm and comic .. It's too short for a movie. Declaration ⁇ 5 Turn around, it's obvious. Declaration
- the target data collected through the data collection server 100 are opinion information data, that is, general sentence / document data and information data given affirmative / negative evaluation thereof, as shown in Table 1 above.
- the positive / negative evaluation may be expressed as a score within a certain range, or may be variously evaluated using an asterisk or other symbols. In the present invention, all of the positive / negative evaluations expressed in this manner are recalculated and used in the same score range.
- the present invention uses a score between 1 and 10 points (positive as closer to 10 points), and the collected data is 2 points when the score is used between 1 and 5 points. If it is, it is calculated as Equation 2 below.
- the collected data as described above is a set of opinion scores ⁇ (data, score), (data, score), (data, score), (data, score) ⁇ converted into corresponding data sentences / documents and scores used in the present invention.
- the web document data collected by the data collection server 100 can be used immediately, it is also possible to apply a domain classification module (not shown) by classifying each domain for use.
- data related to a corresponding domain is collected by each domain (eg, a movie, a book, an electronic product, a cosmetic, a clothing, a person, etc.) determined to be classified to secure data for each domain.
- each domain eg, a movie, a book, an electronic product, a cosmetic, a clothing, a person, etc.
- the data collected for each domain is composed of a combination of review data and fact data for the domain.
- the ratio of opinion data and fact data of the data collected by each domain is maintained at the same or similar ratio, so that the data is purely classified according to domains.
- language processing is performed to extract an appropriate feature from each domain.
- the language processing is divided into semantically separable units through, for example, Morpheme Analyze or Segmentation.
- the characteristics of the corresponding domain are as follows.
- Trigram The author is a book, is a book A, from book A, interesting from A, interesting, interesting, make writing, writing a writing, make up a writing, Was ,.
- the quality of the domain is as follows. That is, as a result of morphological analysis, after removing the investigation, affix, pre-end ending, and ending ending which do not have a special meaning, the qualities in the form of unigram, bigram, and trigram as shown in the above spacing are removed. Can be used.
- Unigram, Bigram, and Trigram features can all be used, or only some of them can be used selectively. This is the case when the evaluation using the evaluation data shows the highest performance. You will select a combination.
- features of each domain are probabilistically learned using, for example, Naive Baysian, SVM or K-NN, and other general machine learning algorithms (Machine Learning Classifier Algorithm).
- linear classifier may be expressed as Equation 4 below.
- the size of the vector is the total number of features, and features that do not appear in the document have "0" values, and features that appear in the document have their number or value "1".
- Is a weight vector which is a vector that gives weight to each feature by each class
- the matrix size is the number of types of features ⁇ class Count
- the machine learning algorithm can use the data in the same manner as described above.
- Naive Baysian as an example, it may be expressed as Equation 5 below.
- C means a class, for example, a domain such as a movie, a book, a product, and the like.
- the F i means each feature, for example, Unigram (author), Bigram (author book), Trigram (author book A), and the like.
- P (C) is the probability that the class C comes out. For example, if there are five movie data, twelve book data, and eight commodity data, P (movie) has a "5 / (5 + 12 + 8)" probability.
- P (F 1 ,..., F n ) is a probability that each F i appears at the same time, and may be omitted because the same applies to all classes (the same applies to all classes as denominators).
- C) is a probability that F 1 ,..., F n is generated when class C is given.
- Equation 5 the calculation of the molecules for determining the actual class probabilities assumes that each feature is conditionally independent of each other, and is specifically calculated as shown in Equation 6 below.
- C) is a probability of F i given a given C
- C) represents the number (Frequency) of the feature (F j ) in any class C.
- the total number of features is N.
- a classification model is generated, and when the sentence or document comes in as an input, the classification model probabilistically determines which domain the data is included in. do.
- the classification model outputs class C showing the highest generation probability for the features.
- a dictionary may be constructed in an automated manner when extracting opinions through the opinion / non-comment classification module 300 described later.
- a learning model for classifying opinion expressions may be automatically generated.
- a model for extracting opinions having optimal performance for a domain may be automatically generated.
- the Internet (10) is a TCP / IP protocol and a number of services existing in the upper layer, that is, Hyper Text Transfer Protocol (HTTP), Telnet, File Transfer Protocol (FTP), Domain Name System (DNS), SMTP (Simple Mail Transfer Protocol), Simple Network Management Protocol (SNMP), Network File Service (NFS), Network Information Service (NIS), and the like, a worldwide open computer network structure that provides the user terminal 900 will be described later It provides an environment that allows easy access to the web server 800. Meanwhile, the Internet 10 may be a wired or wireless internet, or may be a core network integrated with a wired public network, a wireless mobile communication network, or a portable internet.
- HTTP Hyper Text Transfer Protocol
- Telnet Telnet
- FTP File Transfer Protocol
- DNS Domain Name System
- SMTP Simple Mail Transfer Protocol
- SNMP Simple Network Management Protocol
- NFS Network File Service
- NIS Network Information Service
- the Internet 10 may be a wired or wireless internet, or may be a core network integrated with a wired public network
- the language processing module 200 separates the web document data collected from the data collection server 100 or stored in the data storage module 150 in sentence units, and performs language processing on each of the separated sentences. Performs the function of extracting features.
- the language processing module 200 may be a sentence unit for general document data (eg, text, Korean, Word, or Excel document) in addition to the web document data collected from the data collection server 100 or stored in the data storage module 150.
- the linguistic features may be extracted by performing a linguistic processing on the separated sentences.
- the general document data has a pre-set opinion and / or non- opinion set in order to implement the opinion / disagreement classification model, that is, the opinion / disagreement classification module 300 to more accurately distinguish whether the data is opinion data or fact data Sentences can be included, thereby effectively supplementing limited web document data.
- the linguistic processing may be performed by, for example, Morpheme Analyze or Segmentation, but in addition to the irradiation process for extracting features (or index words), Korean refractive processing, or circular return processing, etc. It may be.
- the opinion / non-discrimination module 300 performs a function of dividing the opinion / non-comment sentences by using linguistic features of each sentence extracted from the language processing module 200.
- the sentences extracted from the language processing module 200 include sentences with opinions, and general sentences without opinions. These sentences may be divided into sentences in which an opinion exists and sentences in which an opinion does not exist using the opinion / non-comment classification module 300.
- the opinion / disagreement classification module 300 can be easily implemented using the conventional machine learning algorithm described above.
- a data set composed of opinions and a data set composed only of fact information are collected. Thereafter, for example, Morpheme Analyze or Segmentation is performed to extract an appropriate linguistic feature.
- the spacing is a process of dividing an input sentence into units having meanings. For example, if the input sentence says "I enjoyed the movie”, the resulting sentence translates to "I enjoyed the movie”.
- the morpheme analysis is a task for finding what part of speech information for each of the divided units. For example, if the input sentence says "I enjoyed the movie”, the result sentence reads "I (CTP1 first person pronoun) + (fjb assistant) movie (CMCN secretary common noun) + (fjco purpose check) (YBDO general verbs) + crab (fmoca auxiliary verb) + (YBDO general verbs) + (fmbtp past tense first ending endings) + da (fmofd flat ending endings).
- learning is performed by selecting a general machine learning algorithm, for example, Naive Baysian, SVM, K-NN, or other model.
- a general machine learning algorithm for example, Naive Baysian, SVM, K-NN, or other model.
- the opinion / non-comment classification model that can distinguish whether the data is opinion data or fact data, that is, the opinion / non-comment classification module 300 may be implemented. .
- the opinion / disagreement classification module 300 configured as described above may be provided and implemented for each of the data for each domain classified through the above-described domain classification model.
- the opinion expression division module 400 performs a function of dividing the language features of the opinion sentences separated from the opinion / non- opinion classification module 300 into positive / negative opinion expressions.
- the opinion expression division module 400 finds a part that is a positive / negative opinion among the input comment sentences and displays the part.
- the opinion / disagreement classification module 300 it is also possible to display a positive / negative expression portion in the input sentence using the opinion expression classification module 400 directly.
- the opinion expression classification module 400 quantifies the degree of affirmation / negativeness of all words such as general self-supporting words and words as well as salmon, and utilizes them as a resource, and finds a machine learning model for finding positive / negative expressions in sentences. Used to generate
- the present invention calculates the positive score and the negative score of each semantic unit and automatically stores them in a separate opinion vocabulary storage module (not shown).
- the input data includes a score indicating a positive degree and sentences / documents belonging to the score as shown below.
- these opinion data are collected through review sites where the user posts positive / negative scores and opinions on the general web.
- the probability of how "best” (CMCN secretary common noun) represents positive / negative and how the word "best (CMCN secretary common noun)" is distributed to each score range (1 to 10).
- the calculation is performed through the following equation (7).
- the w j shown below is the "highest (CMCN non-ordinary common noun)", it can represent a combination of words and tag information (POS-Part Of Speech), or can represent a single word except the "best" tag information. have.
- S means all score sets. For example, if a movie evaluation has 1 to 10 points, it means a set of sentences with a score of 1 to 10 points.
- the score (s i ) means the actual score of the corresponding score set. In other words, the score (s i ) of the 10-point score set is 10.
- the score (w i ) represents the positive / negative score of w i .
- Freq (w j , s i ) represents the number of times the word w j appears in the score set s i . Is the sum of the number of occurrences of the word w j in all score sets, which means the number of times w j appears in the entire data.
- Equation 7 For example, assuming that only two 10-point sentences and two 9-point sentences exist as "fun”, it can be obtained as shown in Equation 8 below.
- the semantic unit may be configured as a semantic unit by tying "fun” with the morpheme "YBDO”, or with only one word of "fun”.
- Equation 10 is required in consideration of the number of data in each score band.
- s i ) is a probability value where w j appears in the s i score set.
- the word is normalized by using a probability value appearing in each score band, thereby solving a problem in which scores are biased according to the size of the score band.
- a positive / negative score of each semantic unit is calculated and stored in a separate opinion vocabulary storage module.
- the opinion expression classification module 400 may be provided and implemented for each of the data for each domain classified through the above-described domain classification model.
- the input sentence is "1 (SGR instructional adjective) movie (CMCN secretary common noun) + (fjb assistant) really (SBO general adverb) funny (YBDO general verb) + (fmbtp past tense Mother) + everything (fmofd flat ending)-10 points,
- a vocabulary of more than a certain score among scores of 1 to 10 points is considered as positive, and vocabulary less than or equal to a specific score is regarded as negative vocabulary.
- the opinion expression division module 400 is implemented using the opinion expression division learning model. That is, the opinion expression classification module 400 finds and marks the part of the opinion in detail when the sentence is input as described above.
- sentence 4 belongs to a set of 1-point sentences, it is certain that sentence 4 is a negative sentence, and using the information of such a negative sentence, all the positive / negative vocabulary in sentence 4 is changed to negative vocabulary. Will be displayed. That is, sentence 4 is displayed as follows.
- the models used for learning are, for example, Hidden Markov Model (HMM), Maximum Entropy Model (ME), Conditional Random Field, Struct Support Vector Machine, or other Machine Learning algorithms.
- HMM Hidden Markov Model
- ME Maximum Entropy Model
- Conditional Random Field Struct Support Vector Machine
- Struct Support Vector Machine or other Machine Learning algorithms.
- the data commonly input in these machine learning algorithm models are (x 1 , y 1 ), ..., (x n , y n ), where x is the meaning unit of "funny (YBDO general verb)" or " Funny “," YBDO general verbs ", etc., and y is a level (Label) that the semantic unit can have, such as” Positive “,” Negative “, and” Neutral “. You can also add other levels, such as "Strength,” which will help you determine affirmations.
- the model desired in the present invention is a model for predicting a level y that is eventually attached to the input data sequence xs.
- the model mentioned above are the front and rear (x i- of x i x i with respect to the particular position 1 , y i-1 ), (x i + 1 , y i + 1 ), before and after its (x i-2 , y i-2 ), (x i + 2 , y i + 2 ), like this
- Continuously expanding surrounding data, as well as other feature-part of speech, capital letter, emoticon, etc. information that exists at that location can also be used together so that the level of y i of x i under certain conditions You will predict if this will work.
- the opinion expression division module 400 is generated.
- the opinion expression division module 400 predicts which level sequence y i is generated for the corresponding data sequence.
- the language processing is performed as follows to selectively perform a spacing or morpheme analysis (Morpheme Analyze), such data is input to the opinion expression classification module 400 as follows: Can be expressed as:
- the input sentence is "4 Lee (SGR-directed noun) movie (CMCN secretary common noun) + (fjb assistant) funny (YBDO common verb) + Ji (fmoca auxiliary conjunction) + (YA auxiliary verb) + (Fmbtp past tense fresh ending) + c (fmofd flat ending ending) +. (G symbol)-1 point ",
- Sentences with affirmative / negative opinions are divided into "4 Lee (SGR-directed adjective) / NEUTRAL movie (CMCN secretary common noun) / NEUTRAL + (fjb assistant) / NEUTRAL funny (YBDO general verb) / NEGATIVE + G (fmoca secondary concatenation) / NEGATIVE No (YA auxiliary language) Of NEGATIVE + NEWRAL + c (fmofd flat ending ending) / NEUTRAL +. (G symbol) / NEUTRAL ".
- the opinion expression classification module 400 for each domain after classifying the opinion data displayed positive / negative portion input to the domain classification module You can also create it.
- the indexing server 500 indexes the opinion information of the corresponding web document so that the opinion information of the corresponding web document is stored in the opinion indexing information storage module 600 according to the linguistic qualities of the opinion sentences separated from the opinion expression classification module 400. Perform.
- the opinion indexing information storage module 600 is a database (DB) of the summary information of the corresponding opinion sentences and linguistic features of the corresponding web document by linguistic qualities of each opinion sentence indexed by the indexing server 500 Perform the stored function.
- DB database
- the positive / negative opinion-expression part is found and displayed using the opinion / non- opinion classification module 300 and the opinion expression classification module 400 with respect to the input data.
- information such as a title, a text, an analyzed text, a creation date, a tag, a URL, image information, and video information may be stored.
- the object may include entity keyword information, object keyword and opinion expression relationship information, or type information to which each entity keyword belongs.
- Type information of object keyword (AA, movie), (BB, movie)
- the type information of the object keyword among the information data may be found by mixing the following two methods.
- the first method is to find out the type information of each entity by obtaining an entity database for each predefined type
- the second method is to search the web using the domain classification module. This is a method of classifying document and sentence domains and finding out what type it is.
- the relation information between the object keyword and the opinion expression information is, for example, using a Korean parser or an SVO analysis method (eg, a verb, a verb and an object analysis) method, and each opinion expression is dependent on an entity. Information is determined and inputted.
- a Korean parser or an SVO analysis method eg, a verb, a verb and an object analysis
- the information data as described above is stored in the opinion indexing information storage module 600 so that the opinion search module 700 may be used later.
- the opinion search module 700 receives the user's specific opinion search keyword and / or type information transmitted through the web server 800, and interoperates with the indexing server 500 or the indexing information storage module 600. Searches for indexing information of a web document related to the specific opinion search keyword and / or type information and transmits the indexed information of the web document to the web server 800 to be transmitted to the corresponding user terminal 900.
- the content transmitted to the web server 800 may be "Keyword: Nom Nom, Type: Positive / Negative / Opinion".
- the "opinion” in the type information is a search result in which both positive and negative opinions are displayed together, and the "positive” is a type in which only positive opinions are output.
- “Negative” is a type that only negative opinions.
- the specific opinion search keyword and type are transmitted to the opinion search module 700
- the specific opinion search keyword and data corresponding to the type are read from the indexing server 500 or the indexing information storage module 600.
- the search results are sent back to the web server 800 by ranking such as the amount or date order.
- the searched result information may include, for example, a title, a link, a corresponding site title, a positive number, a negative number, a positive number, a body content, a body summary content, a positive expression position, a negative expression position, and the like.
- the summary content refers to a part of the document in which a part appearing in the searched result document corresponding to the keyword “nom nom” and a part of a positive / negative opinion expression are displayed together.
- the summary section instead of displaying only the search keywords in the body summary content (Snippet) like the general search, the summary section also displays the part where the opinion about the keyword appears.
- the information related to the specific search keyword may be selected through an advertisement selection module (not shown) in which advertisement related data is input by the advertiser in advance, and may be displayed together with the search result.
- the web server 800 receives the specific opinion search keyword and / or type information transmitted from the user terminal 900 connected through the Internet 10, and transmits it to the opinion search module 700, and receives an opinion. It receives the opinion search result, that is, the indexing information data retrieved from the search module 700 performs an interface (Interface) to be displayed on the screen of the user terminal 900.
- an interface Interface
- the opinion search module 700 and the web server 800 are separated from each other, but are not limited thereto, and the opinion search module 700 is integrated into the web server 800 so that the web server is integrated. It may be implemented to perform all functions at (800).
- the web server 800 may display on the screen of the corresponding user terminal 900 to selectively check the overall opinions and positive / negative opinions related to the specific opinion search keyword (FIGS. 3 to 6). Reference).
- the web server 800 may determine the ratio of positive / negative opinion expression in the overall opinion search result associated with the specific opinion search keyword, or the positive / negative opinion expression in each opinion information related to the specific opinion search keyword.
- the ratio may be displayed on the screen of the corresponding user terminal 900 (see FIGS. 3 to 6).
- the web server 800 may list the opinion search results related to the specific opinion search keyword in order of importance or time order (latest order or oldest order) and display them on the screen of the corresponding user terminal 900. have.
- the importance level calculates a ratio of the importance of the specific opinion search keyword in the web document and how many opinions the web document includes.
- the degree of relevance and opinion expression determine the importance.
- the importance may be calculated over the entire time range, or may be limited to a specific time range and applied only to documents in that time zone.
- the time sequence is a method of displaying the web document in ascending / descending order according to the order in which the web document is generated. You can display the entire time in ascending / descending order, or you can show it in chronological order within a specific time range.
- the web server 800 corresponds to a predetermined opinion input window (not shown) so that not only the opinions of other users related to the specific opinion search keyword can be searched, but also their opinions can be added to the searched opinion results in the form of comments. It may be displayed on the screen of the user terminal 900.
- the user can log in or comment in a non-logged state.
- the user inputs gender / age / region and other personal information at the time of membership registration, and statistical information according to gender / age / region and other classifications for opinion information added in this system using this personal information. Can be obtained, which can be provided to other users in a euro / free manner.
- the web server 800 may display a comment search result related to the specific opinion search keyword in a specific expression (eg, underline, By emphasizing in bold text or other colors such as various colors that can be emphasized on the web) and displaying them on the screen of the corresponding user terminal 900, the user's opinion can be more easily distinguished (FIGS. 3 to 6). Reference).
- a comment search result related to the specific opinion search keyword eg, underline, By emphasizing in bold text or other colors such as various colors that can be emphasized on the web
- a comment search result related to the specific opinion search keyword eg, underline, By emphasizing in bold text or other colors such as various colors that can be emphasized on the web
- the web server 800 analyzes the positive / negative opinion part of the opinion search result articles related to the specific opinion search keyword according to the user's selection, and emphasizes this in a specific expression to display the screen of the user terminal 900. Can be displayed (see FIG. 5).
- the web server 800 analyzes the comment for the comment search result article. After performing the operation, the display is displayed on the screen of the corresponding user terminal 900. At this time, the part expressed by the opinion / positive / negative is emphasized to the user with a specific color, a scratched letter, an underline, etc., which can be emphasized on the web.
- the web server 800 may display the results of the opinion search related to the specific opinion search keyword on the screen of the corresponding user terminal 900 in the form of a graph of the change of the positive / negative ratio according to the degree of positive / negative opinion expression. Can be.
- the web server 800 provides statistical analysis data for a specific opinion search keyword input by the user.
- the X axis represents time
- the Y axis represents positive / negative opinion expression (positive / negative).
- a specific opinion search keyword may be displayed as a graph, or a positive / negative ratio change for other specific opinion search keywords belonging to the same category as the specific opinion search keyword may be expressed as a graph.
- the above date information should be stored together in the indexing information storage module 600. Then, the following operation is performed to configure the screen.
- one cycle is selected for each period (day / week / month / year) to find the number of documents in which a particular opinion search keyword is determined to be positive and the number of documents determined to be negative for each period.
- the web server 800 may display a positive / negative ratio of the opinion search result related to the specific opinion search keyword on the screen of the corresponding user terminal 900 for each detailed item of the specific opinion search keyword.
- the sub-item is divided into sub-themes of a corresponding keyword such as sound quality, design, portability, and the like. Can be displayed.
- the web server 800 may display on the screen of the corresponding user terminal 900 to select the user's consent / objection to the opinion search results articles related to the specific opinion search keyword (see FIG. 6). .
- the user can agree or disagree with the opinion search result. This may be reflected by clicking (selecting) the yes / no button on the opinion search result screen as shown in FIG. 6 to be described later.
- the number of votes approved by the user-the number of votes against the user is given as a weight to the ranking of each opinion search result. The more you give the effect of lowering the ranking.
- the pros give the recommendation by the recommendation (w i ) to distribute the profits in the aforementioned advertising platform to benefit the content providers who have received the pros.
- agree (w i ) agree (w i ) -disagree (w i )
- agree (w i ) means the number the user agrees with
- disagree (w i ) means the number the user disagrees with.
- the web server 800 may monitor generation of positive / negative opinions related to a specific opinion search keyword registered by the user in real time and notify the corresponding user terminal 900.
- the monitoring means that the user is notified when a positive / negative opinion related to a specific opinion search keyword registered in advance is notified to the corresponding user, and each company monitors the negative opinion about the company and responds immediately. There is an effect that can be done.
- the web server 800 may display an advertisement associated with the specific opinion search keyword on the screen of the user terminal 900 on the screen where the user inputs a specific opinion search keyword to check the opinion search result for the specific opinion search keyword. Can display
- the order of advertisement placement may be the order in which the advertising billing amount is large, or the corresponding keyword and the relation information. Accordingly, the user can selectively perform general opinion search (positive and negative mixed) / positive opinion search / negative opinion search, and the above-mentioned advertisements are displayed together for each opinion search.
- documents that are positively expressed for each advertisement product may be extracted and provided together with each advertisement in a general online advertisement publication. This will show a positive feedback document extracted with all the available advertising methods online, such as general keyword search ads, opinion search ads, or general banner ads.
- an advertisement product of the corresponding category may be displayed as a search advertisement.
- the number of positive / negative opinions of each product and positive opinions for each product may also be shown.
- advertisers can post their own ads for negative feedback. At this time, it is possible to post clarification articles on general advertisements or corresponding opinions, and at the same time, it is possible to send transcripts of clarification comments on negative comments in a batch.
- the data input by the advertiser may be set by inputting the following data, for example.
- Contents of advertisement Set advertisement link, advertisement text, advertisement image, etc.
- Opinion search keyword Advertisers set their ads to be posted when any opinion search keyword is entered. For example, if the user sets the opinion search keyword "shine phone” and the user inputs the search term "shine phone”, the advertisement of the advertiser who inputs "shine phone” appears.
- the opinion search results are arranged at the top, the order of placement is in accordance with the order of the amount paid by the advertiser.
- posts of users who have positively reviewed the advertisement product together with the advertisement may be posted together.
- Opinion search result keywords Advertisers can set their own ads to be posted when the keyword appears in the opinion search results set in the opinion search results.
- the advertisement of the corresponding advertiser may be posted when "JM53" appears in the opinion search result. This can maximize the advertising effect.
- the advertisement posting position can be placed at the top of the opinion search results or placed together with the opinion search results, and the advertiser can select which opinion search results to send the advertisement to, and search for general / opinion search / positive opinion search / You can choose from negative feedback search results.
- advertising revenue can be shared with a certain percentage of the publisher. Using this, you can set up ads for your own products if you have a positive post about your product, or post a negative for your competitor's product.
- Advertisers can post ads within the analysis page even if they select one of the comment search results to see a page that specifically analyzes the positive / negative part of the comment search result body. have.
- advertisers can post ads selectively based on whether they are mostly positive or negative within the analyzed page, and whether the advertiser's keywords are positive / negative or not is a certain distance between the positive / negative expressions and the keywords entered by the advertiser. It can be determined by how many more positive / negative expressions appear within the distance.
- the data input by the advertiser is the same as the above-described data input content
- the contents input by the website administrators providing the contents of the opinion search result are, for example, name, social security number, account number, site address, Address and so on.
- the user when the user performs the opinion search, for example, the user inputs the opinion search keyword "A" into the search box. Thereafter, the opinion search result is displayed on the screen of the corresponding user terminal 900.
- the top N opinion search result content providers (the corresponding sites) and share the opinion search keyword advertising revenue.
- the content provider sharing the revenue is the target of previously inputting the site information to the search site.
- the amount of distribution of the revenue is given to each weight as follows, based on the proportion of the total, the opinion search keyword advertising revenue is shared.
- the content provider restricts the target to the top N contents of the opinion search result.
- the advertising revenue generated by inputting a single opinion search keyword is "C”
- the proportion of the platform provider that is, the opinion search service provider (search company)
- the opinion search result content provider If the ratio of the profits to be obtained is "1- ⁇ ”, the importance w i of each content provider in the revenue distribution is calculated as in Equation 12 below.
- the registered (w i ) function is a function indicating whether the w i content provider is registered with the opinion search service provider.
- the rank i is a value indicating a search rank in which the content of the w i content provider appears, and has a value of 1 in the case of the first content.
- the rank_weight is a function for determining how much importance is assigned to the opinion search result, and the higher the value, the higher the importance of ranking of the opinion search result is reflected.
- the click (w i ) is a function indicating whether a user who has searched for the corresponding content search result has clicked. Indicates.
- the click_weight is a constant that determines how much weight to give to whether the user clicks.
- the recommendation (w i ) indicates the number of times that users recommend the content.
- the recommended number of times may be two types of recommendation times: a general recommendation number and a recommendation number related to a specific opinion search keyword.
- the recommend_weight represents a weight given to the number of recommendations.
- Equation 12 When Equation 12 is used, a larger portion of revenue is distributed in the case where a user, a site that is clicked more frequently, and a content recommended by more users appear at the top of the opinion search result among registered users.
- Equation 13 the advertisement fee (C) provided by advertisers for each opinion search keyword-specific opinion search result is distributed as shown in Equation 13 below.
- C ⁇ ⁇ is the revenue that the opinion search service provider (search company) takes
- C ⁇ (1- ⁇ ) is the revenue that the content providers bring
- the user terminal 900 is connected to the web server 800 through a wired or wireless communication network such as, for example, a network or the Internet, and is connected to the web server 800 through a conventional web browser.
- a wired or wireless communication network such as, for example, a network or the Internet
- Various services to be provided can be provided.
- a computer such as a desktop PC or a notebook PC is generally, but is not limited thereto, and may be any type of wired / wireless communication device that can access a web server 800 through the Internet 10 and use a bidirectional opinion search service. .
- the user terminal 900 may be a cellular phone, a PCS phone (PCS phone), a synchronous / asynchronous IMT-2000 (International Mobile Telecommunication-) communication via a wireless Internet or a portable Internet.
- PCS phone PCS phone
- IMT-2000 International Mobile Telecommunication-2000
- a mobile terminal such as 2000
- PDA Palm Personal Computer
- PDA Personal Digital Assistant
- WAP phone Smart Phone
- WAP phone Wireless application protocol phone
- Mobile game machine It may mean all wired and wireless home appliances / communication devices having a user interface for accessing a web server 800 that operates a opinion search service such as mobile play-station.
- FIG. 2 is a flowchart illustrating a method of searching for opinions using the Internet according to an embodiment of the present invention
- FIGS. 3 to 6 are screen configuration diagrams for explaining the result of opinion search applied to an embodiment of the present invention.
- 3 is a screen diagram showing a result of opinion search when selecting a specific opinion search keyword (nomnomnom) and a positive opinion type
- FIG. 4 is a opinion search result when selecting a specific opinion search keyword (nomnomnom) and a negative opinion type
- FIG. 5 is a screen configuration diagram illustrating a page function of opinion analysis result for a comment search result article related to a specific comment search keyword (nomnomnom)
- FIG. 6 is a specific comment search keyword ( (Nom Nom Nom) related to the search results, the screen configuration that allows the user to select the pros / cons against the article.
- language processing module 200 By separating the web document data collected in step S100 into sentence units, and performing linguistic processing (eg, morphological analysis or spacing) on each of the separated sentences, language features are extracted (S200). ).
- linguistic processing eg, morphological analysis or spacing
- step S300 after dividing the opinion / non-comment sentences using the linguistic qualities of each sentence extracted in the step S200 through the opinion / disagreement classification module 300 (S300), the opinion expression classification module 400 In step S300, the linguistic qualities of the divided opinion sentences are divided into positive / negative opinion expressions (S400).
- the indexing server 500 performs indexing such that opinion information of the corresponding web document is stored in the opinion indexing information storage module 600 for each language feature of the opinion sentence divided in the step S400 (S500). ).
- the summary information of the corresponding opinion sentence for each language sentence of the opinion sentence indexed in step S500 and the basic and opinion information of the corresponding web document are made into a database and stored in a separate opinion indexing information storage module 600. This is preferred.
- a user who wants to search for opinions accesses a specific web page (eg, http://buzzni.com) that provides a opinion search service using the user terminal 900 capable of accessing the Internet 10.
- the server 800 provides a main search screen having a search input window A for comment search and a type selection button B for selecting a comment search type (comment / positive / negative).
- the web server 800 After receiving a specific opinion search keyword and / or opinion search type transmitted from the user terminal 900 connected through the Internet 10 and delivering it to the opinion search module 700, the opinion search module 700 is a web server 800.
- the opinion search module 700 is a web server 800.
- the indexing server 500 or the opinion indexing information storage module 600 the opinion information of the web document related to the specific opinion search keyword received through) is searched, and the opinion search result is transmitted back to the web server 800.
- the web server 800 displays the opinion search results for the specific opinion search keyword searched through the opinion search module 700 on the screen of the corresponding user terminal 900 (S600).
- the opinion search result related to the specific opinion search keyword is displayed on the screen of the user terminal 900 in step S600, the positive / negative opinion expression in the overall opinion search result related to the specific opinion search keyword is displayed. It is preferable to display the ratio or the ratio of positive / negative opinion expression in each opinion information related to the specific opinion search keyword (see FIGS. 3 to 6).
- step S600 when the opinion search result related to the specific opinion search keyword is displayed on the screen of the corresponding user terminal 900, it is preferable to display the results in order of importance or time.
- the importance is determined based on the degree of relevance and opinion expression that the specific opinion search keyword has in the corresponding web document, and is limited and applied to the entire time range or the specific time range, and the time order is the order in which the corresponding web document is generated. According to the ascending / descending order, it can be applied to the entire time range or limited to a specific time range.
- the corresponding opinion in the form of a comment about the opinion contents of the web document related to the specific opinion search keyword. It is preferable to display a comment input window (not shown) so that a search user's comment can be added.
- the expression expressed as positive / negative together with the specific opinion search keyword is specified (eg, Underlining, bold text, or various colors) is preferably displayed (see FIGS. 3 to 6).
- the opinion analysis page may provide a function of “comment opinion analysis” (FIG. 3 to 6).
- the web server 800 analyzes the positive / negative comment section for the comment search result article, for example, an underscore. It is preferable to highlight the display in at least one of bold font and various colors (see FIG. 5).
- step S600 when the opinion search result related to the specific opinion search keyword is displayed on the screen of the corresponding user terminal 900, the change in the positive / negative ratio of each time according to the degree of positive / negative opinion expression in the form of a graph. Display display is preferred (see FIGS. 3-6).
- the positive / negative ratio may be displayed for each detailed item of the specific opinion search keyword.
- step S600 when the opinion search result related to the specific opinion search keyword is displayed on the screen of the corresponding user terminal 900 in step S600, the user's consent / opposition to the opinion search result articles related to the specific opinion search keyword It is preferable to display on the screen of the corresponding user terminal 900 to select (see FIG. 6).
- the step of monitoring the generation of affirmative / negative opinion related to a specific opinion search keyword registered by the user through the web server 800 in real time to notify the corresponding user terminal 900 may further include.
- the opinion retrieval method using the Internet can also be implemented as computer readable code on a computer readable recording medium.
- the computer-readable recording medium includes all kinds of recording devices in which data that can be read by a computer system is stored.
- a computer-readable recording medium may be a ROM, a RAM, a CD-ROM, a magnetic tape, a hard disk, a floppy disk, a removable storage device, a nonvolatile memory (Flash memory).
- Flash memory nonvolatile memory
- the computer readable recording medium can also be distributed over computer systems connected over a computer network so that the computer readable code is stored and executed in a distributed fashion.
- one embodiment of the present invention implements the opinion retrieval system and method using the Internet based on Korean
- the present invention is not limited thereto, and may be implemented by applying various languages such as English, Japanese, and Chinese.
- FIG. 7 is a block diagram illustrating an overall opinion search and advertisement service system using the Internet according to an embodiment of the present invention.
- opinion information DB 100 may be used.
- advertisement information DB 200 may be used.
- advertisement search module 300 may be used.
- advertisement search module may be used.
- the web server 500 may be used.
- the opinion information DB 100 performs a function of storing opinion information of the corresponding web document as a database (DB) for each language feature of the opinion sentence. That is, in the opinion information DB 100, it is preferable that the summary information of the corresponding opinion sentence for each language and the basic and opinion information of the corresponding web document are stored in a database (DB).
- DB database
- the basic and opinion information of the web document may include, for example, a title, a text, an analyzed text, a creation date, a tag, a URL, an image, a video, the number of positive / negative expressions, the overall positive / negative degree, and each positive / negative object.
- Location information at the beginning and end of the expression, object keyword information to be the object of the opinion vocabulary, object keyword and opinion expression relationship information, or type information to which each object keyword belongs may be composed of at least one information.
- information such as a title, a text, an analyzed text, a creation date, a tag, a URL, image information, and video information may be stored.
- the object may include entity keyword information, object keyword and opinion expression relationship information, or type information to which each entity keyword belongs.
- Type information of object keyword (AA, movie), (BB, movie)
- the type information of the object keyword among the information data may be found by mixing the following two methods.
- the first method is to find the type information of each entity by obtaining an entity database for each predefined type
- the second method is to use a domain classification module (not shown). This is a method of classifying the web document and sentence domain to find out what type it is.
- the relation information between the object keyword and the opinion expression information is, for example, using a Korean parser or an SVO analysis method (eg, a verb, a verb and an object analysis) method, and each opinion expression is dependent on an entity. Information is determined and inputted. The above information data is stored in the opinion information DB 100 so that the opinion search module 300 can be used later.
- a Korean parser or an SVO analysis method eg, a verb, a verb and an object analysis
- the opinion information stored in the opinion information DB 100 is separated into sentence units for web document data existing on the Internet, and language processing is performed on each of the separated sentences to extract linguistic features.
- Opinion / disagreement sentences are classified using the linguistic qualities of the extracted sentences, and the linguistic qualities of the divided opinion sentences are divided into positive / negative opinion expressions, and the linguistic qualities of the divided opinion sentences.
- opinion information of the corresponding web document may be indexed and stored.
- Patent Application No. 2008-93125 (Opinion Retrieval System and Method Using Internet), which was previously filed by the present applicant, describes the opinion information stored in the opinion information DB 100 in detail. Detailed description thereof will be omitted.
- the advertisement information DB 200 performs a function of storing the advertisement information for each keyword into a database (DB). That is, in the advertisement information DB 200, advertisement information for each posting area is stored as a database DB according to the setting of the advertiser.
- the advertisement information may be a database (DB) that is searched by the advertiser and at least one of the advertisement link, advertisement phrase, or advertisement image information for each keyword for the keyword or opinion search types. Preferably stored.
- DB database
- opinion search types may be made of any one type selected from among all opinion contents, positive / negative opinion contents, or positive / negative opinion portion analysis contents of opinion search result articles.
- the data input by the corresponding advertiser through the advertiser terminal 700 may be set, for example, by inputting the following data.
- Contents of advertisement Set advertisement link, advertisement text, advertisement image, etc.
- Opinion search keyword Advertisers set their ads to be posted when any opinion search keyword is entered. For example, if the user sets the opinion search keyword "shine phone” and the user inputs the opinion search word "shine phone”, the advertisement of the advertiser who inputs the "shine phone” comes out.
- the advertisement content is placed on the top of the opinion search results, the order of placement is in accordance with the order of the amount paid by the advertiser.
- posts of users who have positively reviewed the advertisement product together with the advertisement may be posted together.
- Opinion search result keywords Advertisers can set their own ads to be posted when the keyword appears in the opinion search results set in the opinion search results.
- the advertisement of the corresponding advertiser may be posted when "JM53" appears in the opinion search result. This can maximize the advertising effect.
- the advertisement posting position can be placed at the top of the opinion search results or placed together with the opinion search results, and the advertiser can select which opinion search results to send the advertisement to, and search for general / opinion search / positive opinion search / One of the negative feedback search results can be selected.
- advertising revenue can be shared with a certain percentage of the publisher. This allows you to set up ads for your own products if you have a positive post about your product, or post a negative for your competitor's product.
- Analyze page keyword Advertiser posts ads within analyzed page even if user selects one of opinion search results and sees page that specifically analyzes positive / negative part about opinion body. You may.
- advertisers can post ads selectively based on whether they are mostly positive or negative within the analyzed page, and whether the advertiser's keywords are positive / negative or not is a certain distance between the positive / negative expressions and the keywords entered by the advertiser. It can be determined by how many more positive / negative expressions appear within the distance.
- the advertisement information data set by each advertiser is stored in a database (DB) in the advertisement information DB 200 through a web server 500 connected to the Internet.
- DB database
- the opinion search module 300 receives the user's specific opinion search keyword and / or type information transmitted through the web server 500, and works in conjunction with the opinion information DB 100 to provide the specific opinion search keyword. And / or search for opinion information of a web document related to type information and transmit the opinion information to the web server 500 to be transmitted to the corresponding user terminal 600.
- the content transmitted by the user terminal 600 to the web server 500 may be "Keyword: Nom Nom, Type: Positive / Negative / Opinion".
- the "opinion” in the type information is a search result in which both positive and negative opinions are displayed together, and the "positive” is a type in which only positive opinions are output.
- “Negative” is a type that only negative opinions.
- the opinion search module 300 When the specific opinion search keyword and type information is transmitted to the opinion search module 300 as described above, the opinion search module 300 reads the data corresponding to the specific opinion search keyword and the corresponding type from the opinion information DB 100. The search results are sent back to the web server 500 by ranking such as the amount of opinion or the date order.
- the searched result information may include, for example, a title, a link, a corresponding site title, a positive number, a negative number, a positive number, a body content, a body summary content, a positive expression position, a negative expression position, and the like.
- the summary content refers to a part of the document in which a part appearing in the searched result document corresponding to the keyword “nom nom” and a part of a positive / negative opinion expression are displayed together.
- the summary section instead of displaying only the search keywords in the body summary content (Snippet) like the general search, the summary section also displays the part where the opinion about the keyword appears.
- the advertisement search module 400 receives the user's specific opinion search keyword and / or type information transmitted through the web server 500, and works in conjunction with the advertisement information DB 200. And / or search for advertisement information related to type information and transmit the information to the web server 500 to be transmitted to the corresponding user terminal 600.
- the advertisement search module 400 interworks with the advertisement information DB 200 to search for an advertisement associated with a specific keyword input through the web server 500 and the corresponding advertisement information of the search result according to a preset posting area. It is transmitted to the web server 500 to be displayed on the screen of the terminal 600.
- the web server 500 receives the specific opinion search keyword and / or type information transmitted from the user terminal 600 connected through the Internet.
- the opinion search module 300, the advertisement search module 400, and the web server 500 are separated from each other, but are not limited thereto.
- the opinion search module 300 and the advertisement search module are not limited thereto.
- 400 may be integrated into the web server 500 to implement all functions in the web server 500.
- the web server 500 may display on the screen of the user terminal 600 to selectively check the overall opinions and positive / negative opinions related to the specific opinion search keyword.
- the web server 500 may determine the ratio of positive / negative opinion expression in the overall opinion search result associated with the specific opinion search keyword, or the positive / negative opinion expression in each opinion information related to the specific opinion search keyword.
- the advertisement information related to the ratio may be displayed on the screen of the corresponding user terminal 600.
- the web server 500 may list the opinion search results related to the specific opinion search keyword in order of importance or time order (latest order or oldest order) and display them on the screen of the corresponding user terminal 600. have.
- the importance level calculates a ratio of the importance of the specific opinion search keyword in the web document and how many opinions the web document includes.
- the degree of relevance and opinion expression determine the importance.
- the importance may be calculated over the entire time range, or may be limited to a specific time range and applied only to documents in that time zone.
- the time sequence is a method of displaying the web document in ascending / descending order according to the order in which the web document is generated. You can display the entire time in ascending / descending order, or you can show it in chronological order within a specific time range.
- the web server 500 not only retrieves the opinions of other users related to the specific opinion search keyword, but also adds a predetermined opinion input window (not shown) so that the user can add his / her opinion in the form of a comment to the searched opinion result. It may be displayed on the screen of the user terminal 600.
- the user can log in or comment in a non-logged state.
- the user inputs gender / age / region and other personal information at the time of membership registration, and statistical information according to gender / age / region and other classifications for opinion information added in this system using this personal information. Can be obtained, which can be provided to other users in a euro / free manner.
- the web server 500 may include a portion of the opinion search result associated with the specific opinion search keyword together with the portion of the specific opinion search keyword included in each opinion search result text as a positive / negative expression (eg, underline, By bold text or various colors such as expressions that can be emphasized on the web) and displaying them on the screen of the corresponding user terminal 600, the user's opinion can be more easily distinguished.
- a positive / negative expression eg, underline, By bold text or various colors such as expressions that can be emphasized on the web
- the web server 500 analyzes the positive / negative opinion part according to the user's selection of the opinion search result articles related to the specific opinion search keyword, highlights it in a specific expression, and also analyzes the positive / negative analysis.
- the advertisement information related to the opinion part may be displayed on the screen of the corresponding user terminal 600.
- the web server 500 analyzes the comment for the comment search result article. And display advertisement information related to the analyzed opinion contents on a screen of the corresponding user terminal 600. At this time, the part expressed by the opinion / positive / negative is emphasized to the user with a specific color, a scratched letter, an underline, etc., which can be emphasized on the web.
- the web server 500 may display the results of the opinion search results related to the specific opinion search keyword on the screen of the corresponding user terminal 600 in a graph form in accordance with the degree of positive / negative opinion expression. Can be.
- the web server 500 provides statistical analysis data for a specific opinion search keyword input by the user.
- the X-axis indicates time and the Y-axis indicates positive / negative opinion expression (positive / negative).
- the positive / negative ratio of each specific opinion search keyword changes for each period.
- a specific opinion search keyword may be displayed as a graph, or a positive / negative ratio change for other specific opinion search keywords belonging to the same category as the specific opinion search keyword may be expressed as a graph.
- the date information as described above should also be stored in the opinion information DB 100. Then, the following operation is performed to configure the screen.
- one cycle is selected for each period (day / week / month / year) to find the number of documents in which a particular opinion search keyword is determined to be positive and the number of documents determined to be negative for each period.
- the web server 500 may display a positive / negative ratio of the opinion search result related to the specific opinion search keyword on the screen of the corresponding user terminal 600 for each detailed item of the specific opinion search keyword.
- the sub-item is divided into sub-themes of a corresponding keyword such as sound quality, design, portability, and the like. Can be displayed.
- the web server 500 may display on the screen of the corresponding user terminal 600 to select the user's consent / objection to the opinion search results articles associated with the specific opinion search keyword.
- the user can agree or disagree with the opinion search result. This can be reflected by clicking (selecting) the yes / no button on the opinion search result screen.
- the number of votes approved by the user-the number of votes against the user is given as a weight to the ranking of each opinion search result. The more you give the effect of lowering the ranking.
- the pros give the recommendation by the recommendation (w i ) to distribute the profits in the aforementioned advertising platform to benefit the content providers who have received the pros.
- agree (w i ) agree (w i ) -disagree (w i )
- agree (w i ) means the number the user agrees with
- disagree (w i ) means the number the user disagrees with.
- the web server 500 may monitor the generation of positive / negative opinions related to a specific opinion search keyword registered by the user in real time and notify the corresponding user terminal 600.
- the monitoring means that the user is notified when a positive / negative opinion related to a specific opinion search keyword registered in advance is notified to the corresponding user, and each company monitors the negative opinion about the company and responds immediately. There is an effect that can be done.
- the web server 500 screens the advertisement information associated with the specific opinion search keyword on the screen of the user terminal 600 on the screen where the user inputs a specific opinion search keyword to check the opinion search result for the specific opinion search keyword. Can be displayed on
- the order of advertisement placement may be the order in which the advertising billing amount is large, or the corresponding keyword and the relation information. Accordingly, the user can selectively perform general opinion search (positive and negative mixed) / positive opinion search / negative opinion search, and the above-mentioned advertisements are displayed together for each opinion search.
- documents that are positively expressed for each advertisement product may be extracted and provided together with each advertisement in a general online advertisement publication. This will show a positive feedback document extracted with all the available advertising methods online, such as general keyword search ads, opinion search ads, or general banner ads.
- an advertisement product of the corresponding category may be displayed as a search advertisement.
- the number of positive / negative opinions of each product and positive opinions for each product may also be shown.
- advertisers can post their own ads for negative feedback. At this time, it is possible to post clarification articles on general advertisements or corresponding opinions, and at the same time, it is possible to send transcripts of clarification comments on negative comments in a batch.
- the web server 500 may provide a part of the advertising revenue to the content provider providing each opinion search result article according to the search ranking of the corresponding content, whether the search user is selected, and the number of recommendations for the corresponding content.
- the data input by the advertiser is the same as the above-described data input content
- the contents input by the website administrators providing the contents of the opinion search result are, for example, name, social security number, account number, site address, Address and so on.
- the user when the user performs the opinion search, for example, the user inputs the opinion search keyword "A" into the search box. Thereafter, the opinion search result is displayed on the screen of the corresponding user terminal 600.
- the top N opinion search result content providers (the corresponding sites) and share the opinion search keyword advertising revenue.
- the content provider sharing the revenue is the target of previously inputting the site information to the search site.
- the amount of distribution of the revenue is given to each weight as follows, based on the proportion of the total, the opinion search keyword advertising revenue is shared.
- the content provider restricts the target to the top N contents of the opinion search result.
- the advertising revenue generated by inputting a single opinion search keyword is "C”
- the proportion of the platform provider that is, the opinion search service provider (search company)
- the opinion search result content provider If the ratio of the profits to be obtained is " 1- ⁇ ", the importance w i of each content provider in the revenue distribution is calculated as in Equation 15 below.
- the registered (w i ) function is a function indicating whether the w i content provider is registered with the opinion search service provider.
- the rank i is a value indicating a search rank in which the content of the w i content provider appears, and has a value of 1 in the case of the first content.
- the rank_weight is a function for determining how much importance is assigned to the opinion search result, and the higher the value, the higher the importance of ranking of the opinion search result is reflected.
- the click (w i ) is a function indicating whether a user who has searched for the corresponding content search result has clicked. Indicates.
- the click_weight is a constant that determines how much weight to give to whether the user clicks.
- the recommendation (w i ) indicates the number of times that users recommend the content.
- the recommended number of times may be two types of recommendation times: a general recommendation number and a recommendation number related to a specific opinion search keyword.
- the recommend_weight represents a weight given to the number of recommendations.
- Equation 15 the registered users appear in the top of the opinion search results, and take up a greater share in revenue distribution when the user, the site that clicks more frequently, and the content recommended by more users.
- Equation 16 the advertisement fee (C) that the advertisers provide for the opinion search result for each opinion search keyword is distributed as shown in Equation 16 below.
- C ⁇ ⁇ is the revenue that the opinion search service provider (search company) takes
- C ⁇ (1- ⁇ ) is the revenue that the content providers bring
- the user terminal 600 and the advertiser terminal 700 are connected to the web server 500 through a wired or wireless communication network such as a network or the Internet, for example, a typical web browser.
- a wired or wireless communication network such as a network or the Internet, for example, a typical web browser.
- a computer such as a desktop PC or a notebook PC is generally, but is not limited thereto, and may be any type of wired / wireless communication device capable of accessing a web server 500 through the Internet and using a bidirectional opinion search service. .
- the user terminal 600 and the advertiser terminal 700 may be a cellular phone, a PCS phone (PCS phone), a synchronous / asynchronous IMT- that communicates through a wireless Internet or a portable Internet.
- a mobile terminal such as 2000 (International Mobile Telecommunication-2000), in addition to a Palm Personal Computer (PDA), a Personal Digital Assistant (PDA), a Smart Phone, a WAP phone (WAP phone)
- PDA Palm Personal Computer
- WAP phone WAP phone
- FIGS. 9 to 12 illustrate results of opinions searching and advertisements applied to an embodiment of the present invention. These are the screen configuration diagrams.
- the opinion information of the corresponding web document is stored in the opinion information DB 100 for each language feature of the opinion sentence (S100), and the advertisement information for each keyword is advertised.
- Information is stored in the DB (200) (S200).
- a user who wants to search for opinions connects to a specific web page (eg, http://buzzni.com) that provides opinion search and advertisement service using the user terminal 600 capable of internet access.
- 500 provides a main search screen having a search input window A for opinion search and advertisement service, and type selection buttons B for selecting a comment search type (opinion / positive / negative).
- the server 500 receives a specific opinion search keyword and / or opinion search type transmitted from the user terminal 600 connected through the Internet, and delivers it to the opinion search module 300 and the advertisement search module 400, and then the opinion search.
- the module 300 and the advertisement search module 400 may include opinion information DB and advertisement information related to the opinion information of the web document related to the specific opinion search keyword received through the web server 500 and related advertisement information DB. Each search is performed at 200 and the opinion search result and advertisement information are transmitted back to the web server 500.
- the web server 500 includes the opinion search result articles for the specific opinion search keyword respectively searched through the opinion search module 300 and the advertisement search module 400, and the advertisement information related thereto with preset reference information (eg, , The advertisement display order or location, etc.) is properly displayed on the screen of the user terminal 600 (S300).
- preset reference information eg, , The advertisement display order or location, etc.
- step S100 the summary information of the corresponding opinion sentence for each linguistic feature of the opinion sentence and the basic and opinion information of the corresponding web document may be converted into a database (DB) in the opinion information DB 100.
- DB database
- step S100 the opinion information stored in the opinion information DB 100 is divided into sentence units of web document data existing on the Internet, and language processing is performed on each of the separated sentences. And classify the opinion / non-comment sentences using the linguistic qualities of the extracted sentences, and classify the linguistic qualities of the divided opinion sentences into positive / negative opinion expressions.
- the opinion information of the web document can be indexed and stored according to the linguistic qualities of the comment sentence.
- step S200 at least one of the advertisement link, advertisement phrase or advertisement image information for each keyword of the search set by the advertiser in the advertisement information DB 200 and the result for each keyword or opinion search type as a result.
- Information can be stored in a database.
- the opinion retrieval types may be, for example, any one type selected from among whole opinion contents, positive / negative opinion contents, or positive / negative opinion portion analysis contents of opinion search result articles.
- step S300 when displaying the advertisement information associated with the opinion search results articles related to the particular keyword on the screen of the user terminal 600, the entire opinion, positive / negative opinion content associated with the specific keyword is selectively selected. Displayed on the screen of the corresponding user terminal 600 so as to be identified, and the ratio of positive / negative opinion expression in all opinion search results related to the specific keyword, or positive / in each opinion information related to the specific keyword The advertisement information related to the negative opinion expression ratio may be displayed on the screen of the corresponding user terminal 600 (see FIGS. 3A to 3D).
- step S300 when advertising information related to the opinion search result articles related to the specific keyword is displayed on the screen of the corresponding user terminal 600, the advertisement information related to the positive opinion contents related to the specific keyword is displayed.
- An input window (not shown) may be displayed on a screen of the corresponding user terminal 600, or a posting text of a corresponding search user may be posted on negative opinion contents of a web document related to the specific keyword. ) Can be displayed on the screen.
- step S300 when the advertisement information related to the opinion search result articles related to the specific keyword is displayed on the screen of the corresponding user terminal 600, the user selects the opinion search result articles related to the specific keyword.
- the positive / negative opinion portion may be analyzed, and advertisement information related to the analyzed opinion portion may be displayed on the screen of the corresponding user terminal 600 (see FIG. 12).
- the method may further include providing a part of the advertising revenue to the content provider providing each opinion search result article according to the search ranking of the corresponding content, whether the search user is selected, and the number of recommendations for the corresponding content. You may.
- the opinion retrieval and advertisement service method using the Internet can also be implemented as computer-readable code on a computer-readable recording medium.
- the computer-readable recording medium includes all kinds of recording devices in which data that can be read by a computer system is stored.
- a computer-readable recording medium may be a ROM, a RAM, a CD-ROM, a magnetic tape, a hard disk, a floppy disk, a removable storage device, a nonvolatile memory (Flash memory).
- Flash memory nonvolatile memory
- the computer readable recording medium can also be distributed over computer systems connected over a computer network so that the computer readable code is stored and executed in a distributed fashion.
- an embodiment of the present invention implements a feedback search and advertisement service system and method using the Internet based on Korean
- the present invention is not limited thereto, and may be implemented by applying various languages such as English, Japanese, and Chinese. It may be.
Landscapes
- Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Finance (AREA)
- Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Information Transfer Between Computers (AREA)
Abstract
Description
표현 | 점수 | 의견내용 |
★★★★★ | 10 | 재미있어 신고 |
★★★★★ | 10 | '똑똑한' 사람들이 살아가는 이야기 신고 |
★★★★★ | 8 | 현명한 사람들의 일상 뜯어고치기! 신고 |
★★★★★ | 9 | 삼촌의 매력에 흠뻑... 신고 |
★★★★★ | 8 | 스마트한 사람들의 이야기가 아닌 평범한 사람들의 이야기 신고 |
★★★★★ | 10 | 연기도 좋고 내용도 잼있고 가슴 훈훈해지는 사랑이야기. 삼촌 너무 매력적이야∼???? 신고 |
★★★★★ | 10 | 정말 감동할만한 이야기이었어요. 신고 |
★★★★★ | 10 | 별 기대 안하고 봤는데, 보는 내내 가슴 따뜻해지는 영화였습니다. 재미도 있고요 신고 |
★★★★ | 6 | 훈훈하고 코믹하고.. 영화 넘 짧은거 같은데.. 근데 진짜 삼촌없음 어쩔뻔???? 신고 |
★★★ | 5 | 돌고돌고돌아 결국은 뻔한 이야기. 신고 |
Claims (27)
- 인터넷 상에 존재하는 웹 문서 데이터를 수집하는 제1 서버;상기 수집된 웹 문서 데이터에 대해 문장 단위로 분리하고, 분리된 각 문장에 대해 언어처리를 수행하여 언어적인 자질들을 추출하는 언어처리모듈;상기 추출된 각 문장의 언어적인 자질들을 이용하여 의견/비의견 문장을 구분하는 의견/비의견 구분모듈;상기 구분된 의견 문장의 언어적인 자질들에 대해 긍정/부정 의견표현으로 구분하는 의견표현 구분모듈;상기 구분된 의견 문장의 언어적인 자질별로 해당 웹 문서의 의견 정보들이 저장되도록 인덱싱하는 제2 서버; 및인터넷을 통해 접속된 사용자 단말로부터 전송되는 특정 키워드를 제공받아 상기 제2 서버와 연동되어 상기 특정 키워드와 관련된 웹 문서의 의견 정보들을 검색하여 해당 사용자 단말의 화면에 의견 검색결과를 디스플레이 해주는 웹 서버를 포함하는 인터넷을 이용한 의견 검색 시스템.
- 제1 항에 있어서,상기 제1 서버를 통해 수집된 웹 문서 데이터에서 필요한 텍스트, 이미지 또는 비디오 정보들 중 적어도 어느 하나의 정보 데이터를 추출하여 저장되는 데이터 저장모듈이 더 포함되는 것을 특징으로 하는 인터넷을 이용한 의견 검색 시스템.
- 제1 항에 있어서,상기 언어처리모듈은 상기 수집된 웹 문서 데이터와 함께 미리 설정된 의견/비의견 문장들을 포함하는 일반적인 문서 데이터에 대해 문장 단위로 분리하고, 분리된 각 문장에 대해 언어처리를 수행하여 언어적인 자질들을 추출하는 것을 특징으로 하는 인터넷을 이용한 의견 검색 시스템.
- 제1 항에 있어서,상기 제2 서버를 통해 인덱싱된 각 의견 문장의 언어적인 자질별 해당 의견 문장의 요약정보 및 해당 웹 문서의 기본 및 의견 정보들이 데이터베이스(DB)화하여 저장되는 의견 인덱싱 정보 저장모듈이 더 포함되는 것을 특징으로 하는 인터넷을 이용한 의견 검색 시스템.
- 제1 항에 있어서,상기 웹 서버는 상기 특정 키워드와 관련된 전체 의견, 긍정적/부정적 의견내용을 선택적으로 확인할 수 있도록 해당 사용자 단말의 화면에 디스플레이 해주거나, 특정 키워드와 관련된 전체적인 의견 검색결과 내에서의 긍정/부정 의견표현의 비율, 또는 상기 특정 키워드와 관련된 각 의견 정보 내에서의 긍정/부정 의견표현의 비율을 해당 사용자 단말의 화면에 디스플레이 해주는 것을 특징으로 하는 인터넷을 이용한 의견 검색 시스템.
- 제1 항에 있어서,상기 웹 서버는 상기 특정 키워드와 관련된 의견 검색결과를 중요도 또는 시간 순서에 따라 리스트(List)화하여 해당 사용자 단말의 화면에 디스플레이 해주되,상기 중요도는 상기 특정 키워드가 해당 웹 문서에서 가지는 관련도와 의견표현 정도를 통해 결정되고, 전체 시간 범위 또는 특정 시간 범위로 제한하여 적용되며,상기 시간 순서는 해당 웹 문서가 생성되는 순서에 따라 오름차순/내림차순으로 결정되고, 전체 시간 범위 또는 특정 시간 범위로 제한하여 적용되는 것을 특징으로 하는 인터넷을 이용한 의견 검색 시스템.
- 제1 항에 있어서,상기 웹 서버는 상기 특정 키워드와 관련된 웹 문서의 의견 내용들에 대해 댓글 형식으로 해당 의견 검색 사용자의 의견을 추가할 수 있도록 의견 입력창을 해당 사용자 단말의 화면에 디스플레이 해주거나, 상기 특정 키워드와 관련된 의견 검색결과를 상기 특정 키워드와 함께 긍정/부정으로 표현된 부분을 특정한 표현으로 강조하여 해당 사용자 단말의 화면에 디스플레이 해주는 것을 특징으로 하는 인터넷을 이용한 의견 검색 시스템.
- 제1 항에 있어서,상기 웹 서버는 상기 특정 키워드와 관련된 의견 검색결과 글들에 대해 해당 사용자의 선택에 따라 긍정/부정 의견 부분을 분석하고, 이를 특정한 표현으로 강조하여 해당 사용자 단말의 화면에 디스플레이 해주는 것을 특징으로 하는 인터넷을 이용한 의견 검색 시스템.
- 제1 항에 있어서,상기 웹 서버는 상기 특정 키워드와 관련된 의견 검색결과를 긍정/부정 의견표현 정도에 따라 시기별로 긍정/부정 비율의 변화를 그래프 형태로 해당 사용자 단말의 화면에 디스플레이 해주거나, 상기 특정 키워드와 관련된 의견 검색결과를 상기 특정 키워드의 세부 항목별로 긍정/부정 비율을 해당 사용자 단말의 화면에 디스플레이 해주는 것을 특징으로 하는 인터넷을 이용한 의견 검색 시스템.
- 제1 항에 있어서,상기 웹 서버는 상기 특정 키워드와 관련된 의견 검색결과 글들에 대해 해당 사용자의 찬성/반대 의사를 선택할 수 있도록 해당 사용자 단말의 화면에 디스플레이 해주거나, 사용자가 기 등록한 특정 키워드와 관련된 긍정/부정 의견의 생성을 실시간으로 모니터링 하여 해당 사용자 단말로 통지해주는 것을 특징으로 하는 인터넷을 이용한 의견 검색 시스템.
- 의견 문장의 언어적인 자질별로 해당 웹 문서의 의견 정보들이 저장되는 의견정보 DB;키워드별 광고 정보들이 저장되는 광고정보 DB; 및인터넷을 통해 접속된 사용자 단말로부터 전송되는 특정 키워드를 제공받아 상기 의견정보 DB 및 광고정보 DB와 연동되어 상기 특정 키워드와 관련된 웹 문서의 의견 및 광고 정보들을 각각 검색하여 해당 사용자 단말의 화면에 의견 검색결과 글들과 함께 관련된 광고 정보를 디스플레이 해주는 웹 서버를 포함하는 인터넷을 이용한 의견 검색 및 광고 서비스 시스템.
- 제11 항에 있어서,상기 의견정보 DB에 저장되는 의견 정보들은,인터넷 상에 존재하는 웹 문서 데이터에 대해 문장 단위로 분리하고, 분리된 각 문장에 대해 언어처리를 수행하여 언어적인 자질들을 추출하고, 상기 추출된 각 문장의 언어적인 자질들을 이용하여 의견/비의견 문장을 구분하며, 상기 구분된 의견 문장의 언어적인 자질들에 대해 긍정/부정 의견표현으로 구분하며, 상기 구분된 의견 문장의 언어적인 자질별로 해당 웹 문서의 의견 정보들을 인덱싱하여 저장되는 것을 특징으로 하는 인터넷을 이용한 의견 검색 및 광고 서비스 시스템.
- 제11 항에 있어서,상기 광고 정보들은 광고주에 의해 기 설정된 검색과 그 결과 키워드별 또는 의견검색 타입들에 대한 결과 키워드별 광고 링크, 광고 문구 또는 광고 이미지 정보 중 적어도 어느 하나의 광고 정보가 데이터베이스(DB)화하여 저장되는 것을 특징으로 하는 인터넷을 이용한 의견 검색 및 광고 서비스 시스템.
- 제11 항에 있어서,상기 웹 서버는 상기 특정 키워드와 관련된 전체 의견, 긍정적/부정적 의견내용을 선택적으로 확인할 수 있도록 해당 사용자 단말의 화면에 디스플레이 해주고, 상기 특정 키워드와 관련된 전체 의견 검색결과 내에서의 긍정/부정 의견표현의 비율, 또는 상기 특정 키워드와 관련된 각 의견 정보 내에서의 긍정/부정 의견표현의 비율과 함께 관련된 광고 정보들을 해당 사용자 단말의 화면에 디스플레이 해주는 것을 특징으로 하는 인터넷을 이용한 의견 검색 및 광고 서비스 시스템.
- 제11 항에 있어서,상기 웹 서버는 각 의견 검색결과 글을 제공하는 컨텐츠 제공자에게 해당 컨텐츠의 검색 순위, 검색 사용자의 선택 여부 및 해당 컨텐츠에 대한 추천 횟수에 따라 광고 수익의 일부를 제공해주는 것을 특징으로 하는 인터넷을 이용한 의견 검색 및 광고 서비스 시스템.
- (a) 인터넷 상에 존재하는 웹 문서 데이터를 수집하는 단계;(b) 상기 수집된 웹 문서 데이터에 대해 문장 단위로 분리하고, 분리된 각 문장에 대해 언어처리를 수행하여 언어적인 자질들을 추출하는 단계;(c) 상기 추출된 각 문장의 언어적인 자질들을 이용하여 의견/비의견 문장을 구분하는 단계;(d) 상기 구분된 의견 문장의 언어적인 자질들에 대해 긍정/부정 의견표현으로 구분하는 단계;(e) 상기 구분된 의견 문장의 언어적인 자질별로 해당 웹 문서의 의견 정보들이 저장되도록 인덱싱하는 단계; 및(f) 인터넷을 통해 접속된 사용자 단말로부터 전송되는 특정 키워드와 관련된 웹 문서의 의견 정보들을 검색하여 해당 사용자 단말의 화면에 의견 검색결과를 디스플레이 해주는 단계를 포함하는 인터넷을 이용한 의견 검색 방법.
- 제16 항에 있어서,상기 단계(b)에서, 상기 수집된 웹 문서 데이터와 함께 미리 설정된 의견/비의견 문장들이 포함된 일반적인 문서 데이터에 대해 문장 단위로 분리하고, 분리된 각 문장에 대해 언어처리를 수행하여 언어적인 자질들을 추출하는 것을 특징으로 하는 인터넷을 이용한 의견 검색 방법.
- 제16 항에 있어서,상기 단계(f)에서, 상기 특정 키워드와 관련된 의견 검색결과를 해당 사용자 단말의 화면에 디스플레이 할 경우, 상기 특정 키워드와 관련된 전체 의견, 긍정적/부정적 의견내용을 선택적으로 확인할 수 있도록 디스플레이 해주거나, 상기 특정 키워드와 관련된 전체적인 의견 검색결과 내에서의 긍정/부정 의견표현의 비율, 또는 상기 특정 키워드와 관련된 각 의견 정보 내에서의 긍정/부정 의견표현의 비율을 디스플레이 해주는 것을 특징으로 하는 인터넷을 이용한 의견 검색 방법.
- 제16 항에 있어서,상기 단계(f)에서, 상기 특정 키워드와 관련된 의견 검색결과를 해당 사용자 단말의 화면에 디스플레이 할 경우, 중요도 또는 시간 순서에 따라 디스플레이 해주되,상기 중요도는 상기 특정 키워드가 해당 웹 문서에서 가지는 관련도와 의견표현 정도를 통해 결정하고, 전체 시간 범위 또는 특정 시간 범위로 제한하여 적용하며,상기 시간 순서는 해당 웹 문서가 생성되는 순서에 따라 오름차순/내림차순으로 결정하고, 전체 시간 범위 또는 특정 시간 범위로 제한하여 적용하는 것을 특징으로 하는 인터넷을 이용한 의견 검색 방법.
- 제16 항에 있어서,상기 단계(f)에서, 상기 특정 키워드와 관련된 의견 검색결과를 해당 사용자 단말의 화면에 디스플레이 할 경우, 상기 특정 키워드와 관련된 웹 문서의 의견 내용들에 대해 댓글 형식으로 해당 의견 검색 사용자의 의견을 추가할 수 있도록 의견 입력창을 디스플레이 해주거나, 상기 특정 키워드와 함께 긍정/부정으로 표현된 부분을 특정한 표현으로 강조하여 디스플레이 해주는 것을 특징으로 하는 인터넷을 이용한 의견 검색 방법.
- 제16 항에 있어서,상기 단계(f)에서, 상기 특정 키워드와 관련된 의견 검색결과를 해당 사용자 단말의 화면에 디스플레이 할 경우, 상기 특정 키워드와 관련된 의견 검색결과 글들에 대해 해당 사용자의 선택에 따라 긍정/부정 의견 부분을 분석한 후, 밑줄, 굵은 글씨체 또는 다양한 색상 중 적어도 어느 하나의 표현으로 강조하여 디스플레이 해주거나, 긍정/부정 의견표현 정도에 따라 시기별 긍정/부정 비율의 변화를 그래프 형태로 디스플레이 해주는 것을 특징으로 하는 인터넷을 이용한 의견 검색 방법.
- (a) 의견 문장의 언어적인 자질별로 해당 웹 문서의 의견 정보들을 별도의 의견정보 DB에 저장하는 단계;(b) 키워드별 광고 정보들을 별도의 광고정보 DB에 저장하는 단계; 및(c) 인터넷을 통해 접속된 사용자 단말로부터 전송되는 특정 키워드와 관련된 웹 문서의 의견 및 광고 정보들을 상기 의견정보 DB 및 광고정보 DB에서 각각 검색하여 해당 사용자 단말의 화면에 의견 검색결과 글들과 함께 관련된 광고 정보들을 디스플레이 해주는 단계를 포함하는 인터넷을 이용한 의견 검색 및 광고 서비스 방법.
- 제22 항에 있어서,상기 단계(a)에서, 상기 의견정보 DB에 저장되는 의견 정보들은,인터넷 상에 존재하는 웹 문서 데이터에 대해 문장 단위로 분리하고, 분리된 각 문장에 대해 언어처리를 수행하여 언어적인 자질들을 추출하며, 상기 추출된 각 문장의 언어적인 자질들을 이용하여 의견/비의견 문장을 구분한 후, 상기 구분된 의견 문장의 언어적인 자질들에 대해 긍정/부정 의견표현으로 구분하며, 상기 구분된 의견 문장의 언어적인 자질별로 해당 웹 문서의 의견 정보들을 인덱싱하여 저장하는 것을 특징으로 하는 인터넷을 이용한 의견 검색 및 광고 서비스 방법.
- 제22 항에 있어서,상기 단계(c)에서, 상기 특정 키워드와 관련된 의견 검색결과 글들과 함께 관련된 광고 정보를 해당 사용자 단말의 화면에 디스플레이 할 경우, 상기 특정 키워드와 관련된 전체 의견, 긍정적/부정적 의견내용을 선택적으로 확인할 수 있도록 해당 사용자 단말의 화면에 디스플레이 해주고, 상기 특정 키워드와 관련된 전체 의견 검색결과 내에서의 긍정/부정 의견표현의 비율, 또는 상기 특정 키워드와 관련된 각 의견 정보 내에서의 긍정/부정 의견표현의 비율과 함께 관련된 광고 정보들을 해당 사용자 단말의 화면에 디스플레이 해주는 것을 특징으로 하는 인터넷을 이용한 의견 검색 및 광고 서비스 방법.
- 제22 항에 있어서,상기 단계(c)에서, 상기 특정 키워드와 관련된 의견 검색결과 글들과 함께 관련된 광고 정보를 해당 사용자 단말의 화면에 디스플레이 할 경우, 상기 특정 키워드와 관련된 긍정 의견 내용들과 함께 관련된 광고 정보들을 해당 사용자 단말의 화면에 디스플레이 해주거나, 상기 특정 키워드와 관련된 웹 문서의 부정 의견 내용들에 대해 해당 검색 사용자의 해명글을 게시할 수 있도록 입력창을 해당 사용자 단말의 화면에 디스플레이 해주는 것을 특징으로 하는 인터넷을 이용한 의견 검색 및 광고 서비스 방법.
- 제22 항에 있어서,상기 단계(c)에서, 상기 특정 키워드와 관련된 의견 검색결과 글들과 함께 관련된 광고 정보를 해당 사용자 단말의 화면에 디스플레이 할 경우, 상기 특정 키워드와 관련된 의견 검색결과 글들에 대해 해당 사용자의 선택에 따라 긍정/부정 의견 부분을 분석하고, 상기 분석된 의견 부분과 함께 관련된 광고 정보를 해당 사용자 단말의 화면에 디스플레이 해주는 것을 특징으로 하는 인터넷을 이용한 의견 검색 및 광고 서비스 방법.
- 제22 항에 있어서,상기 단계(c)이후에, 각 의견 검색결과 글을 제공하는 컨텐츠 제공자에게 해당 컨텐츠의 검색 순위, 검색 사용자의 선택 여부 및 해당 컨텐츠에 대한 추천 횟수에 따라 광고 수익의 일부를 제공해주는 단계를 더 포함하는 것을 특징으로 하는 인터넷을 이용한 의견 검색 및 광고 서비스 방법.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US13/120,649 US20110179009A1 (en) | 2008-09-23 | 2009-09-23 | Internet-based opinion search system and method, and internet-based opinion search and advertising service system and method |
Applications Claiming Priority (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
KR1020080093125A KR101007284B1 (ko) | 2008-09-23 | 2008-09-23 | 인터넷을 이용한 의견 검색 시스템 및 그 방법 |
KR10-2008-0093125 | 2008-09-23 | ||
KR10-2008-0096415 | 2008-10-01 | ||
KR1020080096415A KR101021204B1 (ko) | 2008-10-01 | 2008-10-01 | 인터넷을 이용한 의견 검색 및 광고 서비스 방법 |
Publications (2)
Publication Number | Publication Date |
---|---|
WO2010036012A2 true WO2010036012A2 (ko) | 2010-04-01 |
WO2010036012A3 WO2010036012A3 (ko) | 2010-07-22 |
Family
ID=42060261
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/KR2009/005405 WO2010036012A2 (ko) | 2008-09-23 | 2009-09-23 | 인터넷을 이용한 의견 검색 시스템, 의견 검색 및 광고 서비스 시스템과 그 방법 |
Country Status (2)
Country | Link |
---|---|
US (1) | US20110179009A1 (ko) |
WO (1) | WO2010036012A2 (ko) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2012044305A1 (en) * | 2010-09-30 | 2012-04-05 | Hewlett-Packard Development Company, L.P. | Identification of events of interest |
Families Citing this family (27)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11023675B1 (en) | 2009-11-03 | 2021-06-01 | Alphasense OY | User interface for use with a search engine for searching financial related documents |
US9576251B2 (en) * | 2009-11-13 | 2017-02-21 | Hewlett Packard Enterprise Development Lp | Method and system for processing web activity data |
US8725773B2 (en) | 2011-08-30 | 2014-05-13 | E-Rewards, Inc. | System and method for generating a knowledge metric using qualitative internet data |
WO2013114754A1 (ja) * | 2012-02-02 | 2013-08-08 | 株式会社コナミデジタルエンタテインメント | 情報提供システム、サーバ装置、記録媒体および制御方法 |
CN104272301B (zh) * | 2012-04-25 | 2018-01-23 | 国际商业机器公司 | 用于提取一部分文本的方法、计算机可读介质和计算机 |
US9092506B1 (en) * | 2012-05-07 | 2015-07-28 | Google Inc. | Providing a report based on a negative keyword |
US20160133160A1 (en) * | 2012-05-07 | 2016-05-12 | Zoll Medical Corporation | Rescue Performance Metric |
US9396179B2 (en) * | 2012-08-30 | 2016-07-19 | Xerox Corporation | Methods and systems for acquiring user related information using natural language processing techniques |
US9715493B2 (en) * | 2012-09-28 | 2017-07-25 | Semeon Analytics Inc. | Method and system for monitoring social media and analyzing text to automate classification of user posts using a facet based relevance assessment model |
US9141686B2 (en) * | 2012-11-08 | 2015-09-22 | Bank Of America Corporation | Risk analysis using unstructured data |
CN103870973B (zh) * | 2012-12-13 | 2017-12-19 | 阿里巴巴集团控股有限公司 | 基于电子信息的关键词提取的信息推送、搜索方法及装置 |
US9436663B2 (en) * | 2012-12-17 | 2016-09-06 | Hewlett Packard Enterprise Development Lp | Presenting documents to a user based on topics and collective opinions expressed in the documents |
WO2014141452A1 (ja) * | 2013-03-14 | 2014-09-18 | 株式会社 東芝 | 文書分析装置及び文書分析プログラム |
JP6044959B2 (ja) * | 2013-07-31 | 2016-12-14 | インターナショナル・ビジネス・マシーンズ・コーポレーションInternational Business Machines Corporation | 調査を支援する装置及び方法 |
US9213730B2 (en) * | 2013-08-13 | 2015-12-15 | Xerox Corporation | Method and apparatus for extracting portions of text from long social media documents |
CN103593786A (zh) * | 2013-08-14 | 2014-02-19 | 北京世纪喜鹊信息科技有限公司 | 基于互联网的广告定向维度投放网络硬件装置 |
CN103559625B (zh) * | 2013-09-18 | 2017-09-05 | 深圳市东信时代信息技术有限公司 | 一种广告效果监测的方法及系统 |
US9569510B2 (en) | 2013-09-30 | 2017-02-14 | International Business Machines Corporation | Crowd-powered self-improving interactive visualanalytics for user-generated opinion data |
CN103631861A (zh) * | 2013-10-28 | 2014-03-12 | 百度在线网络技术(北京)有限公司 | 一种用于处理并提供评测信息的方法与设备 |
US20150286627A1 (en) * | 2014-04-03 | 2015-10-08 | Adobe Systems Incorporated | Contextual sentiment text analysis |
US10949753B2 (en) | 2014-04-03 | 2021-03-16 | Adobe Inc. | Causal modeling and attribution |
EP3206135A4 (en) * | 2014-10-06 | 2018-06-13 | Hitachi, Ltd. | Sentence retrieval method and sentence retrieval system |
US10929264B2 (en) * | 2016-09-14 | 2021-02-23 | International Business Machines Corporation | Measuring effective utilization of a service practitioner for ticket resolution via a wearable device |
KR102582544B1 (ko) | 2018-05-03 | 2023-09-25 | 삼성전자주식회사 | 전자 장치 및 그 동작 방법 |
US11928709B2 (en) * | 2018-11-19 | 2024-03-12 | Walmart Apollo, Llc | Method and apparatus for automatically providing advertisements |
US20200234331A1 (en) * | 2019-01-17 | 2020-07-23 | Michael Sadowsky | System and process to estimate persuasiveness of public messaging using surveys |
WO2024085806A1 (en) * | 2022-10-19 | 2024-04-25 | Grabtaxi Holdings Pte. Ltd. | Server and method for processing consumer reviews |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2001209728A (ja) * | 1999-11-15 | 2001-08-03 | Power To The People:Kk | 商品等評価データ集計装置、商品等評価データ集計方法、商品等評価装置、商品等評価方法および記録媒体 |
KR20040053369A (ko) * | 2001-11-26 | 2004-06-23 | 후지쯔 가부시끼가이샤 | 정보 해석 방법 및 장치 |
JP2006146567A (ja) * | 2004-11-19 | 2006-06-08 | Internatl Business Mach Corp <Ibm> | 表現検出システム、表現検出方法、及びプログラム |
JP3917629B2 (ja) * | 2000-08-21 | 2007-05-23 | 富士通株式会社 | 自然文処理装置及び自然文処理用プログラム |
Family Cites Families (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
NZ520663A (en) * | 2000-02-10 | 2004-05-28 | Involve Technology Inc | System for creating and maintaining a database of information utilizing user defined keyword relevance ratings |
EP1276061A1 (en) * | 2001-07-09 | 2003-01-15 | Accenture | Computer based system and method of determining a satisfaction index of a text |
US8200477B2 (en) * | 2003-10-22 | 2012-06-12 | International Business Machines Corporation | Method and system for extracting opinions from text documents |
US7962461B2 (en) * | 2004-12-14 | 2011-06-14 | Google Inc. | Method and system for finding and aggregating reviews for a product |
US7548929B2 (en) * | 2005-07-29 | 2009-06-16 | Yahoo! Inc. | System and method for determining semantically related terms |
US20070192166A1 (en) * | 2006-02-15 | 2007-08-16 | Leviathan Entertainment, Llc | Survey-Based Qualification of Keyword Searches |
JP4870448B2 (ja) * | 2006-02-23 | 2012-02-08 | 株式会社日立製作所 | 情報処理装置、顧客ニーズ分析方法、およびプログラム |
US8069182B2 (en) * | 2006-04-24 | 2011-11-29 | Working Research, Inc. | Relevancy-based domain classification |
WO2007140364A2 (en) * | 2006-05-26 | 2007-12-06 | Monitor110, Inc. | Method for scoring changes to a webpage |
US8209320B2 (en) * | 2006-06-09 | 2012-06-26 | Ebay Inc. | System and method for keyword extraction |
US8332269B2 (en) * | 2006-06-27 | 2012-12-11 | Adchemy, Inc. | System and method for generating target bids for advertisement group keywords |
US8296168B2 (en) * | 2006-09-13 | 2012-10-23 | University Of Maryland | System and method for analysis of an opinion expressed in documents with regard to a particular topic |
US7761287B2 (en) * | 2006-10-23 | 2010-07-20 | Microsoft Corporation | Inferring opinions based on learned probabilities |
US20080215571A1 (en) * | 2007-03-01 | 2008-09-04 | Microsoft Corporation | Product review search |
US8200663B2 (en) * | 2007-04-25 | 2012-06-12 | Chacha Search, Inc. | Method and system for improvement of relevance of search results |
US7987188B2 (en) * | 2007-08-23 | 2011-07-26 | Google Inc. | Domain-specific sentiment classification |
US8239189B2 (en) * | 2008-02-26 | 2012-08-07 | Siemens Enterprise Communications Gmbh & Co. Kg | Method and system for estimating a sentiment for an entity |
-
2009
- 2009-09-23 WO PCT/KR2009/005405 patent/WO2010036012A2/ko active Application Filing
- 2009-09-23 US US13/120,649 patent/US20110179009A1/en not_active Abandoned
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2001209728A (ja) * | 1999-11-15 | 2001-08-03 | Power To The People:Kk | 商品等評価データ集計装置、商品等評価データ集計方法、商品等評価装置、商品等評価方法および記録媒体 |
JP3917629B2 (ja) * | 2000-08-21 | 2007-05-23 | 富士通株式会社 | 自然文処理装置及び自然文処理用プログラム |
KR20040053369A (ko) * | 2001-11-26 | 2004-06-23 | 후지쯔 가부시끼가이샤 | 정보 해석 방법 및 장치 |
JP2006146567A (ja) * | 2004-11-19 | 2006-06-08 | Internatl Business Mach Corp <Ibm> | 表現検出システム、表現検出方法、及びプログラム |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2012044305A1 (en) * | 2010-09-30 | 2012-04-05 | Hewlett-Packard Development Company, L.P. | Identification of events of interest |
Also Published As
Publication number | Publication date |
---|---|
WO2010036012A3 (ko) | 2010-07-22 |
US20110179009A1 (en) | 2011-07-21 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2010036012A2 (ko) | 인터넷을 이용한 의견 검색 시스템, 의견 검색 및 광고 서비스 시스템과 그 방법 | |
WO2010068068A2 (ko) | 사용자의 의도에 기반한 정보 검색방법 및 정보 제공방법 | |
WO2010036013A2 (ko) | 웹 문서에서의 의견 추출 및 분석 장치 및 그 방법 | |
WO2017209564A1 (ko) | 앱 리스트 제공 방법 및 그 장치 | |
WO2012074338A2 (ko) | 자연어 및 수학식 처리 방법과 그를 위한 장치 | |
WO2017176100A1 (en) | Method and device for translating object information and acquiring derivative information | |
WO2020017849A1 (en) | Electronic device and method for providing artificial intelligence services based on pre-gathered conversations | |
WO2012134180A2 (ko) | 문장에 내재한 감정 분석을 위한 감정 분류 방법 및 컨텍스트 정보를 이용한 다중 문장으로부터의 감정 분류 방법 | |
WO2015170945A1 (ko) | 영어 어순 지도를 이용한 영어 학습방법 및 그 시스템 | |
WO2011007935A1 (ko) | 홈페이지 통합 서비스 제공 시스템 및 방법 | |
WO2019027240A1 (en) | ELECTRONIC DEVICE AND METHOD FOR PROVIDING A RESEARCH RESULT THEREOF | |
WO2020009297A1 (ko) | 도메인 추출기반의 언어 이해 성능 향상장치및 성능 향상방법 | |
WO2020044290A1 (ko) | 특허 문서 작성 장치, 방법, 컴퓨터 프로그램, 컴퓨터로 판독 가능한 기록매체, 서버 및 시스템 | |
WO2012091360A2 (ko) | 유저 맞춤형 컨텐츠 제공 방법 및 시스템 | |
WO2014021567A1 (ko) | 메시지 서비스 제공 방법, 이를 위한 장치 및 시스템 | |
WO2020222623A9 (ko) | 기획 판매를 위한 컨텐츠를 자동적으로 구축하는 시스템 및 그 방법 | |
WO2019177182A1 (ko) | 속성 정보 분석을 통한 멀티미디어 컨텐츠 검색장치 및 검색방법 | |
WO2018117685A1 (en) | System and method of providing to-do list of user | |
WO2009134096A2 (ko) | 광고노출 제어시스템 및 광고노출 제어방법 | |
WO2018034426A1 (ko) | 커널 rdr을 이용한 태깅 말뭉치 오류 자동수정방법 | |
WO2022065811A1 (en) | Multimodal translation method, apparatus, electronic device and computer-readable storage medium | |
WO2020197257A1 (ko) | 가시적 표현 요소를 이용한 번역 방법 및 그 장치 | |
WO2021137637A1 (en) | Server, client device, and operation methods thereof for training natural language understanding model | |
WO2021215804A1 (ko) | 대화형 청중 시뮬레이션을 제공하는 장치 및 방법 | |
EP3818492A1 (en) | Communication via simulated user |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 09816405 Country of ref document: EP Kind code of ref document: A2 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
WWE | Wipo information: entry into national phase |
Ref document number: 13120649 Country of ref document: US |
|
32PN | Ep: public notification in the ep bulletin as address of the adressee cannot be established |
Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 14/07/2011) |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 09816405 Country of ref document: EP Kind code of ref document: A2 |