KR101513660B1 - Historical information retrieval system based on period query - Google Patents

Abstract

The present invention relates to a history information retrieval system based on a periodic query in which history information is retrieved based on a period query reflecting a time attribute and intuitive information acquisition is enabled. A history information management unit for processing and storing historical information collected on the basis of a period query for retrieving contents of information based on a change of the history information; A history information structuring unit for extracting and clustering relationships between historical objects retrieved from the history information management unit according to the period query, And a history information providing unit for generating an input query in the history information management unit and providing the structured history object information in the history information structuring unit.

Description

[0002] Historical information retrieval system based on period query [

The present invention relates to a history information retrieval system, and more particularly, to a history information retrieval system based on a periodic query in which history information is retrieved based on a period query reflecting a time attribute and intuitive information is acquired.

Recently, due to the rapid development of the Internet and the web, a huge amount of web documents have been created and accumulated on the web. Among these web documents, users generally search for information through a search engine to search for desired information. Currently, the Internet search engine, typified by Google, takes a representative keyword for the information that the user wants to search as a query, and simply takes the web documents including the keywords in order. However, the documents provided as a result of the search may be conflicting or various contents, such as the topic or the opinion expressed by the document, even if the keyword is included or related to the keyword. This change in content is caused by the description of the contents of the web document based on the temporal background. In describing the same subject due to the time attribute contained in the information content, contents such as public opinion, fashion, It can be found in many documents that changes with background or temporal flow.

For example, if you conduct a search on the subject of the Seoul Bus System, the contents of a web document such as a blog or a newspaper article created before or at the beginning of the project are often inconvenient or unnecessary policies . However, documents written after one or two years after the implementation have many advantages such as transfer or exclusive lane, which is convenient or useful. Since the information content of a web document implies a time attribute, the content of the web document varies depending on a specific time period or a viewpoint. However, in the present invention, an information providing method that collectively lists related documents based on a keyword index provided by a search engine It is difficult to sufficiently reflect the time attribute of such information. In order to solve this problem, if we look at the search results provided through keyword index-based information retrieval from the viewpoint of the time attribute of the web document by different viewpoints, It is just a collection of all relevant web documents. Therefore, a user searching for information about a specific time period or a certain time point, such as before or after the invasion of Japan, or searching for information contents changed over time, should perform repetitive information search and complex filtering process using only keywords , Which requires users to spend time and effort inefficiently.

However, if the change of the information contents based on the time axis can be retrieved by reflecting the time attribute of the web document instead of the keyword index, it is possible to obtain new types of information such as the situation before and after the Japanese invasion, Search will be possible. In addition, since this search reflects the time attribute of the web document, it allows the user to access the contents of the information actually sought more accurately, and in particular, It can help you to detect changes. In particular, if the search can be performed by selecting the time zone as a time unit such as 1 year or 10 years, that is, by setting a search interval such as 1 year and 1 year before the FTA, You will be able to search for changes in content. In addition, it is easy to search for relevant key information such as a time point corresponding to an inflection point at which the contents of information about a search word change.

Time can be considered as an important dimension in information space and can be very useful in various fields of information retrieval such as information extraction, topic retrieval, and questionanswering . In particular, it is possible to reconstruct the result of information retrieval using temporal information, or to reflect the temporal information in the calculation of the retrieval result ranking, thereby providing information retrieved by the user as a more accurate retrieval result (see Non-Patent Document 1) . However, in the information retrieval system including the conventional search engine, information retrieval that does not utilize the advantage of temporal information as much as possible is being performed.

However, in recent years, there has been an intriguing research on recognizing the importance of temporal information in information retrieval and analyzing and pioneering it (see Non-Patent Document 2).

In order to utilize temporal information in the information retrieval field, it is necessary to extract the temporal information indicated in the web document. In general, the temporal information of a Web document can be extracted simply by information such as the date of creation of the document or the date of the last modification. However, in order to use accurate time information for the purpose of information retrieval, it is necessary to extract not only meta information of a simple document but also temporal information existing in the contents of the document. For example, J. Pustejovsky (see Non-Patent Document 3) extracts a temporal expression in a context from a proposal of XML-based TimeML in order to extract temporal information existing on the content of a document, Time, Duration, and Set. In addition, a portion corresponding to a temporal expression in a sentence is normalized to a standard form representing real time, i.e., 'YYYY-MM-DD', so that a normalized expression of temporal information .

However, since the temporal representation in actual documents occurs not only in the above four types but also in various forms, the normalization is difficult to perform simply. O.Alonso (refer to non-patent document 4) divides temporal information forms appearing in web documents into three types in order to accurately extract such temporal expressions that occur in various ways. The types of expressions that can be expressed are explicit expressions that correspond directly with time such as 'December 2004', Implicit expressions that are named with specific names such as anniversaries or events, such as 'Labor Day 2008' We defined it as a relational expression that can understand the temporal information through comparison with the data. By defining these three types, not only the temporal information of the metadata of the document but also the temporal expression existing in the context can be extracted more accurately.

This process of temporal information extraction and normalization is called temporal tagging, and the machine that performs time tagging is called a temporal tagger. Time staggers preprocess sentences through a basic text processing process such as morphological analysis and a series of natural language processing and use machine learning techniques in preprocessed sentences to check the boundaries of temporal expressions, Extracted and normalized (see Non-Patent Document 5).

The temporal information extracted by this method is utilized in various fields of information retrieval. M. Gertz (see Non-Patent Document 6) extracts temporal information of web documents provided as a result of information retrieval and performs clustering on the basis of the extracted information, thereby reconstructing search results by clustering web documents having similar time information Respectively. In addition, J. Makkonen (see Non-Patent Document 7) has used a similar study to extract the temporal information of each article by using news articles that have relatively clear temporal information, and to measure the similarity between articles Respectively. In another aspect of providing search results, temporal information can be utilized in the creation of snippets or summaries of documents provided as search results. O.Alonso (refer to non-patent document 8) extracts time information of the metadata of the document and time information on the text and considers this information together with the subject of the document, and has conducted research to generate a part of the most suitable text as a snippet.

In order to extract useful information as well as utilization in providing search results, research has been conducted on data mining using time information from a blog (see Non-Patent Document 9). In addition, studies are being conducted to detect the subject of a document and to express the detected subject on a related time line (see Non-Patent Document 10), and to combine with the Future Retrieval field to predict the relation between some facts or objects in the future (See Non-Patent Document 11), which utilize temporal information, are used in various information retrieval fields. In addition, there is a temporal pattern (see non-patent document 12) that analyzes or recognizes the characteristics and patterns of information using recent temporal information, a time information of the contents of a web document, In the field of Fresh Information Retrieval (refer to non-patent document 13) used for searching, it is possible to map information in accordance with the order of time on geographic information, or to use Spatio-Temporal time information is used as an important factor in research related to information retrieval such as exploration (refer to non-patent document 14).

Meanwhile, the information retrieval technology has been rapidly developing with the development of the Internet since the late 90's. In recent years, the amount of web documents has been rapidly increased, and a large amount of document indexing technology has been used. There is a need for an effective search ranking technique. In particular, information retrieval in the area of the web is focused on the point that information of various fields are connected to each other and quickly and accurately finds the technology development. Currently, Google (www.google.com), a representative web search engine, provides a service that obtains a high search effect from a user's point of view by introducing various search and ranking algorithms such as a ranking system called "Page Rank" have. In addition to the information retrieval technology, various researches such as elimination of duplicate retrieval result, meta retrieval, and specialized retrieval have been actively conducted (refer to non-patent document 15).

In another aspect of information retrieval, a semantic web that provides a well-defined meaning to information on the web and provides a search result suitable for the user's query intention by interpreting the meaning of the computer, Research on knowledge - based information retrieval such as query - based retrieval service is being actively carried out. These knowledge-based information retrieval services not only provide related documents to the user's query but also provide the retrieval result by processing the fragmented information into a single knowledge so that the user can acquire information about the query more easily, It is meaningful that you can acquire information as knowledge as information.

The intellectual service of Naver (www.naver.com) is a representative question-and-answer type knowledge search service. The intellectual services are open-minded because everyone can ask and answer any topic, and because of the communicative nature of sharing the accumulated data, Businessweek has paid attention to intellectual services because it needs to learn from NHN. And has become an area of information retrieval (see Non-Patent Document 16). Although unreliable answers or speculative answers may reduce the reliability, accuracy, and expertise of answers and result in a decrease in the quality of knowledge, it is possible to answer the user's query accurately, And that it is easy to obtain relevant information from the accumulated answers.

In the knowledge search service including the intellectual service, the query is inputted into the search window of the search engine, and the knowledge search system searches the knowledge database for documents likely to be suitable for the user's query. Therefore, it is most important to select a document suitable for the search query from the knowledge database. Although a considerable amount of research has been conducted on the quality evaluation of general web documents, researches on knowledge search services such as question-and-answer services are currently underway. Bivens-Tatum (see Non-Patent Document 17) introduced the history and the status of the knowledge search service, and J. Hill (see Non-Patent Document 18) performed a simple performance test on the expert query response service. Park (refer to non-patent reference 19) performed experiments such as response rate and response time for Naver, Yahoo, and Empas knowledge search services, and compared the performances. Through online surveys, And information usage behavior, and suggested ways to improve the knowledge search service. In addition, Choi (see Non-Patent Document 20) compares and evaluates the topic expression performance of the clustering technique and the manual subject classification for the query response documents of Naver intellectuals.

There is a semantic search method using an ontology that allows a computer to automatically search for a meaningful document by grasping the meaning of the document by accessing from other aspects of the knowledge-based information retrieval. Until now, most of the information on the web has been configured to be easy to read and interpret by people, not the computer. Therefore, it is difficult for the computer to intelligently interpret and process the information. A general information retrieval process of a user is as follows. A user searches for desired information using a search engine, checks the information with the user's eyes, selects necessary information, and reinterprets the selected information. If a user has a search engine that interprets desired information, extracts only core information, processes it, and provides it to the user's intention, the user will be able to escape repetitive and exhausting tasks. In this sense, the Semantic Web documents are meant to be easily interpreted by computers, which makes it possible to search through high level knowledge using the meanings of automated agents or sophisticated search engines.

Assigning a semantic to the web means not to add a natural language processing function for a user interface but to give declarative additional information to facilitate processing by the computer (refer to non-patent document 21). Research on ontologies related to the definition of knowledge for semantic web as a research on semantic web and researches such as RDF and RDFS for describing web resources (see non-patent reference 22), and automated And research on an autonomous program agent (see Non-Patent Document 23).

The ontology defines the concepts of the domain and the interrelationships between these concepts. In the knowledge-based system, it refers to a knowledge system that is unambiguously defined and agreed upon by a wide range of people to facilitate processing by a computer program (Non-Patent Document 24 Reference). Using this well-defined knowledge system, we can classify documents by hierarchically defining the relationship between themes and themes, and search suitable for user's intention. XML and RDF are used to define these ontologies. XML specifies an arbitrary document structure using tags defined by the user, and RDF represents metadata related to Web documents such as title, author, and last modification date, which are bases for representing a Web document. RDF is basically a subject, a verb (or a descriptor) and an object of a general sentence. By defining a pair of these three pieces of information, a specific object such as a person or a web document, Represents a state having a specific value. Through such a method, it is possible to search for information that reflects meaning by classifying documents using meaning. DAML + OIL (refer to non-patent document 25), which is a combination of DAML (DARPA Agent Markup Language) and OIL (Ontology Inference Layer) developed mainly in Europe, based on RDF, OWL (see Non-Patent Document 26) defined by W3C.

Meanwhile, another technique proposed in the prior art for solving the disadvantages of the keyword search is a method of extracting the cultural information described in the patent document 1, Korean Patent Laid-Open Publication No. 10-2011-0041320 (published on April 21, 2011) And related search methods and systems using time / space / subject classification.

Korean Patent Publication No. 10-2011-0041320A, 2011. 04. 21, pp. 8-9.

J.F. Allen. Maintaining Knowledge about Temporal Intervals. In Communications of the ACM, 26 (11): 832-843, 1983 O. Alonso, M. Gertz, and R. Baeza-Yaters, On the value of temporal information in information retrieval. SIGIR Forum, 41 (2): 35-41,2007. J. Pustejovsky, J. M. Castaño, et al. TimeML: Robust Specification of Event and Temporal Expressions in Text. In Proceedings of the AAAI Spring Symposium on New Directions in Question Answering, pages 28-34, 2008. O.Alonso, et. al. Temporal information retrieval: Challenges and opportunities. In International Temporal Web Analytics Workshop, pages 1-8, 2011 O. Kolomiyets and M.-F. Moens. Meeting TempEval-2: Shallow Approach for Temporal Tagger. In Proceedings of the Workshop on Semantic Evaluations: Recent Achievements and Future Directions (SEW '09), pages 52-57, 2009. O. Alonso, M. Gertz, and R. Baeza-Yates. Clustering and Exploring Search Results Using Timeline Constructions. In Proceedings of the 18th ACM International Conference on Information and Knowledge Management (CIKM '09), pages 97-106, 2009. J. Makkonen and H. Ahonen-Myka. Utilizing Temporal Information in Topic Detection and Tracking. In Proceedings of 7th European Conference on Research and Advanced Technology for Digital Libraries (ECDL '03), pages 393-404, 2003. O. Alonso, R. Baeza-Yates, and M. Gertz. Effectiveness of Temporal Snippets. In Proceedings of the Workshop on Web Search Result Summarization and Presentation (WSSP 09), pages 1-4, 2009. A. Qamra, B. Tseng, and E. Chang. Mining Blog Stories Using Community-based and Temporal Clustering. In Proceedings of the 15th ACM International Conference on Information and Knowledge Management (CIKM '06), pages 58-67, 2006. R. Swan and J. Allan. TimeMine: Visualizing Automatically Constructed Timelines. In Proceedings of the 23rd Annual International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR '00), page 393, 2000. A. Jatowt, K. Kanazawa, S. Oyama, and K. Tanaka. Supporting Analysis of Future-related Information in the News Archives and the Web. In Proceedings of the 9th Joint Conference on Digital Libraries (JCDL '09), 2009. B. Shaparenko, R. Caruana, J. Gehrke, and T. Joachims. Identifying Temporal Patterns and Key Players in Document Collections. In Proceedings of the IEEE ICDM Workshop on Temporal Data Mining: Algorithms, Theory and Applications (TDM '05), pages 165-174, 2005. Toyoda, M., & Kitsuregawa, M. What's Really New on the Web? Identifying New Pages from a Series of Unstable Web Snapshots. In WWW2006: Proceedings of the 15th International World Wide Web Conference (pp. 233-241). Edinburgh, Scotland. May 23-26: ACM Press. J. Strotgen, M. Gertz, and P. Popov. Extraction and Exploration of Spatio-temporal Information in Documents. In Proceedings of the 6th Workshop on Geographic Information Retrieval, pages 1-8, 2010. Monika Henzinger, "Google Tutorial: Web Information Retrieval ", Tutorial on Web Information Retrieval at ICDE'2000 (16th International Conference on Data Engineering), 2000. L. Moon and E. Woyke, 2006 "NHN: the little search engine that could" businessweek. 30 January Bivens-Tatum, R. W. 2001. "Expert services on the Web: the commercial compettion for libraries", College & Research Libraries News. 62 (7): 714-716. J. Hill, A. Rolfe. 2001, "Ask-an-expert services analysis" Journal of the American Society for Information Science and Technology. 52 (13): 1106-1121. P.Joo-Bum, J. Dong-Youl. 2004. "An Empirical Study on Web-based Question-Answer Services", In Proceedings of the Korea Society for Information Management, 21 (3): 83-98, 2004. C. sang-hee. 2005. "A Study on Clustering Query - Answer Documents with Structural Features ", In Proceedings of the Korea Society for Library and Information Science, 39 (4): 105-118, 2005. S.Y. Park, J.H. Lee, J.W. Jeon, 2006, "Evaluation of the documents from the Web-based Question and Answer Service ", In Proceedings of the Korea Society for Library and Information Science, 40 (2): 299-314, 2006. Klein, M., "XML, RDF, and Relatives," IEEE Intelligent Systems, vol.16, no.2, March / April, 2001, pp. 26-28. James Hendler, "Agents and the Semantic Web," IEEE Intelligent Systems, vol. 16, no. 2, March / April, 2001, pp. 30-37. T. R. Gruber, "A Translation Approach to Portable Ontologies," Knowledge Acquisition, 5 (2): 199-220, 1993. Debora L. McGuinness, Richard Fikes, James Hendler and Lynn Andrea Stein, "DAML + OIL: An Ontology Language for the Semantic Web," IEEE Intelligent Systems, vol.17, no.5, September / October, 2002, pp.72- 80. Mike Dean et al. Web Ontology Language (OWL) Reference Version 1.0, W3C Working Draft 12 November, 2002, http://www.w3.org/TR/owl-ref/. Kang, Seung-Shik, "Word Classification and Data Structure for Korean Morphological Analysis", The 8th Annual Conference on Human & Cognitive Language Technology, 1996, pp241-245

However, in the conventional techniques described above, it is necessary to input a keyword of a target to be searched. However, since it is difficult to specify a search query by keyword, the history information such as 'the 16th century shipbuilding era' .

And history information has a limit to provide the same search result for the same keyword because the history information search service can not cope with such change because the relationship between person and event changes according to the passage of time.

An object of the present invention is to provide a history information retrieval system based on a periodic query in which history information is retrieved based on a period query reflecting a time attribute and intuitive information acquisition is possible.

Another object of the present invention is to provide a history information retrieval system based on a periodic query that can be searched for a specific time period corresponding to a time index.

It is still another object of the present invention to provide a history information retrieval system based on a periodic query that enables observation of a change in information content over time.

It is still another object of the present invention to provide a historical query-based historical information system capable of extracting and clustering association information between historical objects with respect to historical information searched for through a period of time to acquire intuitive information of a user, And a search system.

According to an aspect of the present invention, there is provided a history information retrieval system based on a periodic query, the history information retrieval system including: a history information collecting unit for retrieving historical information through a search engine, A history information management unit for processing and storing historical information collected on a query basis; A history information structuring unit for extracting and clustering relationships between historical objects retrieved from the history information management unit according to the period query, And a history information providing unit for generating an input query to the history information management unit and providing the structured history object information in the history information structuring unit.

In the above, the history information management unit is configured to convert the processed history object and the annotation information into a time index database.

The history information management unit includes a data collection unit for collecting web page information and old document information including a history object from a search engine; A data processing unit processing the history information and the old document information collected by the data collecting unit according to periods and sorting the frequency of the history objects; And a data server unit for converting the history object and annoy information processed by the data processor into a time index database.

In the above, the historical object is the name of the person who refers to the existential person described in the history, the cultural property which is a worthy tangible / intangible asset left from the past, the event that is the beginning or center of history development, And a place name.

The history information structuring unit includes a history object extracting unit for extracting a history object of the web page and the old document information retrieved from the history information managing unit, an association degree determining unit for determining a degree of association between the extracted history objects, And a detailed relationship judging section for judging the detailed relationship.

The input query includes a period query for defining a period for searching and a keyword query for specifying a keyword of the history object to be searched.

The history information providing unit includes a search input unit for inputting an input query of a target to be searched, a query generating unit for generating an input query input from the search input unit and transmitting the generated input query to the history information management unit, And a history information display unit.

According to the present invention, the collected history information is converted into a database through a time-based index, and the search for the data search is performed over a period of time, thereby minimizing the process for acquiring history information of the user, , The relationship between various historical objects can be obtained based on the period.

In addition, according to the present invention, there is an effect that more efficient search can be performed by focusing on the content characteristic of information and sorting the data by using the characteristic as index reference.

Further, according to the present invention, the search result is provided as a graph-like interface that can be grasped intuitively by humans, thereby providing an advantage that relevant information can be grasped at a glance.

In addition, according to the present invention, it is possible to provide various retrieval access methods through a history object composed of one or more of life, cultural property, event, and place name, and it is also possible to search for two or more historical object relations.

In addition, according to the present invention, the accuracy of the historical information can be improved by aligning the web page information and the old document information according to the morphological analysis and the frequency of occurrence.

Brief Description of the Drawings Fig. 1 is a block diagram of a history information retrieval system based on periodic queries according to the present invention;
FIG. 2 is a conceptual diagram showing a detailed operation of the history information management unit of FIG. 1;
FIGS. 3A to 3C are diagrams showing examples of the history information stored in the history information management unit of FIG. 1,
FIG. 4 is a first conceptual diagram showing a detailed operation of the history information structuring unit of FIG. 1,
FIG. 5 is a second conceptual diagram showing the detailed operation of the history information structuring unit of FIG. 1;
6A and 6B are diagrams showing an example of the information extraction process in the history information structuring unit of FIG. 1,
7A to 7D are diagrams showing the frequency of a history object and the result of soft determination between historical objects in the present invention,
FIG. 8 is a view showing an example of the history information providing unit of FIG. 1,
9A to 9D illustrate examples of search results of history information in the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A history query system based on a periodic query according to a preferred embodiment of the present invention will now be described in detail with reference to the accompanying drawings.

1 is a block diagram of a history information search system based on a periodic query according to a preferred embodiment of the present invention.

As shown in FIG. 1, a history query system based on a periodic query according to the present invention includes a history information manager 100, a history information structure unit 200, and a history information provider 300.

The history information management unit 100 collects historical information through a search engine and processes and stores historical information collected on the basis of a period query for searching for information contents based on a change in time And includes a data collecting unit 110, a data processing unit 120, and a data server unit 130.

Here, the search engine can be classified into a directory search engine, an index search engine, a meta search engine, and the like, and includes an integrated search engine that integrates the results of the above-mentioned directory, index, and meta search engine according to classification of expression .

In general, most of the search engines follow a method of providing keywords as search queries and matching documents as search results, and as a result, they do not deviate much from indexing keywords.

In order to solve the disadvantages of the keyword search as described above, the present invention focuses on retrieving the contents of information centered on the change of time, and in order to do so, a time-based indexing method suitable for performing search based on time- to provide. When indexing a time other than a keyword, it is possible to search for a desired time period instantaneously by using time as an information search query, and thereby it is possible to easily retrieve the information of the corresponding time period information. Also, by comparing search results based on a certain time period, it is possible to easily search for changes in information over time.

The present invention will now be described by way of an example of a time index for historical information for time-based retrieval. In the case of general web documents, since the date of creation of the document and the contents of the document do not coincide with each other in time, the time indicated by the metadata of the document can not be replaced with the time that the contents of the document contain. In other words, except for special cases such as newspaper articles, it is common that the date of the document is different from the time of the document. However, in the case of historical information written in an elaborate manner, the date of creation of the document and the contents of the document coincide with each other in a timely manner. Therefore, indexing based on time is reasonable in terms of contents and is easy to index. Using the advantage of this index, we collected the souce of the Joseon Dynasty from the implementation of the actual system, and performed a time - based index based on day, month, and year. In this way, a period query, which is an information retrieval query, is generated in the form of a time-lapse range, so that information on the range of the corresponding time period can be extracted from the database.

For the definition of historical information to be used in the present invention, the dictionary definition of the word "history" is defined as "a history of human being, a fact occurring in human behavior, or a record of the fact".

To clearly use the meaning of historical information, historical information is defined from the above definition of history. The definition of 'history' has two characteristics in common. One is the fact that it is recorded around some time in the past, and the other is that it is described in some object such as human being, things, events. In other words, it means a fact or record of an object that existed at a certain time in the past. In the present invention, the history information is defined as "information about an object corresponding to a specific time in the past" and this definition is used in the future development process.

The historical information thus defined has two characteristics in terms of technology. First, most historical information is written around time. Although there are many ways to describe historical information, such as the entire history and articles, there are many historical records such as 'Goryeo Byeonyo', 'Dongukguk', and 'Chosun Dynasty Records' . The present invention focuses on the description method of historical information recorded in order of time and takes a method of indexing based on time rather than indexing based on existing keywords. By indexing on the basis of time in this manner, it is possible to search for information by time period in correspondence with the Period Query which will be described later, thereby making it easy to observe the temporal change of the historical information. As a second characteristic of historical information, history information is described with related objects, and the relationship between the described objects changes with the passage of time.

For example, in a particular year, it can be seen that related persons such as Namgon and Zhao Kwang are described in association with each other. Through this technique, history information can be confirmed that objects such as persons and events are described together. Also, in 1517, it can be considered that the relationship between the two is so intense that it is enough for the Namgong to hold office in the office. However, in 1520, it can be confirmed that the King Namgon was ordered to kill the king. Thus, in 1520, the relationship between Gwanggyo and the South can be considered very hostile.

Historical information, like this example, is characterized not only by establishing the relationship between each person, thing, and event, but also by changing the relationship between history information based on time change. Therefore, it is not easy to search for the above characteristics by simple keyword search, and it is not until after a lot of information filtering process that the above-mentioned changing relationship can be grasped.

Through the characteristics of historical information, it can be confirmed that the history information is described together with related history information. The objects that are related to each other in this way are used as queries in keyword search through the search engine, and they are the main object to obtain information mainly. In this way, the object which is searched for by history information is characterized in that the type is limited to a specific range. For example, 'Korean Historical Information Integration System', which is widely used in public, provides historical information through directory classification. In general, directory classification is performed mostly for each type of search. In the case of the directory classification of 'Korean Historical Information Integration System', it is categorized as' person, map, relic, book, research data ', and it can be classified as' person (60%), %), Map, place name (10%) and others (5%). In addition, the content index classification of 'Korean National Culture Encyclopedia', which is a popular history encyclopedia, is classified into 'life, place name, signature, potential (event, office)'. In view of the directory classification or contents index of such popular information providing systems, it has been found that the historical information is mainly described and searched around 'people, cultural properties, place names and events'. Through this fact, Historical information in a broad sense is used as a 'historical object'.

The period query newly generated in the present invention is generated by simply selecting a certain period of time, and is an information search query based on a search-based time included in history information rather than a keyword-type query.

As described above, the history information is described around time. From this fact, it is easier to search for the desired history information if the history information can be searched using the time which is the central element of the technology rather than the search of the history information by the keyword. The periodic query used in the present invention is suitable for retrieving historical information described around a time since a search for a certain time or period is performed, that is, a search for a time zone is performed. In addition, by performing a temporal search by generating a query based on a continuous time with respect to historical information having a feature in which the contents of information or the relationship between objects change in accordance with a change of time, Making it easy to acquire information. In addition, a history is selected for a time-based history information that can not be represented by a keyword such as "a major person relationship in the 1630s" There are advantages to be able to.

In the present invention, since the history information retrieval system converts the history information into a time index database, the retrieval of the history information of the corresponding time period is performed based on the proposed Period Query

And constructs search results for the information retrieved by the query and provides it to the user. This query can be generated by selecting the time domain on the interface of the implemented system. It can be generated not only by itself, but also by performing additional time search on the result retrieved by keyword search in combination with existing keyword search method. have.

In the present invention, as an example, the Chosun Dynasty (1394-1911) was taken as a basic data for the collection of historical information of Joseon Dynasty. The Chosun Dynasty annals are data suitable for constructing a time index as a description method of the elders. The configuration of the system is largely divided into the following three parts as shown in Fig. First, it collects and processes the historical information corresponding to the Joseon Dynasty, and collects and indexes historical information to store it according to the time index. Second, it searches the information of the relevant city in response to the Period Query, And a history information providing section which provides a graphical search result and provides it on the system interface.

In particular, the system interface of the last history information provision part compares with the search result of the existing search engine, and it is confirmed that the system proposed by the present invention can search and acquire efficient historical information reflecting the time factor as compared with the existing search engine .

In order to retrieve historical information corresponding to a user's query, it is necessary to collect historical information in advance, and to process and index it. The history information collects digitized Chosun Dynasty annals by producing a wrapper 112, and the collected annotations are classified and text preprocessed in order to correspond to the period query and the keyword search.

To this end, the data collection unit 110 collects web page information 11 and old document information 13 including a history object through a search engine. The data collecting unit 110 includes a crawler 111 for collecting keywords of history objects and collecting web page information 11 including the keywords and a wrapper 112 for collecting old document information 13 ). Here, the historical object refers to a person who refers to an existential person described in history, a cultural property that is a valuable or intangible asset left from the past, an event that is the beginning or center of history development, It is preferable to include at least one or more of them.

In order to maintain the credibility in the collection of historical objects, we collect mainly the terms derived from the process of nationalization and digitization of historical materials at national level. The keywords are collected from public domain sites such as 'Korea History Compilation Committee, Korea Classical Translation Institute, and Korean Historical Information Integration System', and they are collected in the categories of history terminology, nominative terminology, human index and signature index according to the definition of historical objects. Remove duplicate and insoluble words in the collected keywords and use about 25,000 as historical objects. Also, it keeps the form of 'Hangul' for collection of historical objects with exact meaning such as distinction of homonyms.

The crawler 111 collects the web page information 11 while moving along the link on the web, and the collected web page information is indexed through various steps in the indexer. Such indexing includes steps such as inverse transformation, N-gram, and Tagging. The indexed data is stored in the database and retrieved by the searcher according to the user's query. FIG. 3A shows an embodiment of a web page collected through the crawler 111. FIG.

The wrapper 112 finds information provided in a predetermined form such as a table form in a web document, and forms a database according to the form. When the wrapper 112 is used for the history of the Joseon Dynasty, as shown in FIG. 3B, an old document such as the Chosun Dynasty Annals of the related data is configured in the form of a table having a constant property, Is effective in information collection. In addition, the old document information is easy to classify into category, unlike general web information.

In addition, the data processor 120 processes the web page information and the old document information collected by the data collecting unit 110 according to a period of time index format using time attributes, removes unnecessary tags or keywords, In order of frequency.

The data processing unit 120 includes an insoluble tag removing unit 121 for removing insoluble tags of the web page information 11 and the old document information 13 collected by the data collecting unit 110, A morphological analysis unit 122 for analyzing morphemes of the web page information 11 and the old document information 13, a keyword removing unit 123 for removing non-historical object keywords of the web page information 11 and the old document information 13, And a history object sorting unit 124 that sorts the history information according to the number of history object frequencies of the page information 11 and the old document information 13. [

As shown in FIG. 3C, the data processing unit 120 extracts only general nouns other than the other morphemes through the morpheme analysis unit 122 for processing the data. In addition, general nouns excluding the historical objects defined in the extracted general nouns are removed, and the frequency of the history objects included in the Chosun dynasty annals page and each history object is calculated and sorted in descending order.

The data server unit 130 classifies the web page information 21 and the old document information 23 processed in the data processing unit 120 into a database based on a time index, And transmits the extracted information to the history information structuring unit 200. 2, the data server unit 130 includes a web page storage unit 131 for storing a web page for each history object, an old document storage unit 132 for storing the old document information, And a frequency net web page storage unit 134 for storing the sorted web pages in order of the frequency of the history objects, as shown in FIG. .

In addition, the history information structuring unit 200 extracts and associates the relationship between the historical objects retrieved from the history information management unit 100 according to the period query, and organizes the structured relationship. 4, the history information structuring unit 200 includes a history object extracting unit 220, an association degree determining unit 230, a detail relation determining unit 240, and a service server unit 250.

However, in the process of retrieving based on the keyword query pointing to the keyword of the history object, the period of the first search period is the entire period of the Joseon Dynasty, but the period is excluded As shown in FIG. 5, it can be confirmed that the period query is used in both the period and keyword search.

The history object extracting unit 220 extracts a history object from the retrieved history information.

If the input query is a period query, it can extract the temporal change of the historical information with the feature that the contents of the information or the relation between the historical objects change according to the period, and the information extraction through the period query is the collected data, And the web page. In the case of the Chosun Dynasty Annals, since the types arranged in the database are already sorted in chronological order, it can be easily extracted through the SQL operator (<,>) as shown in FIG. 6A have.

In addition, since web page information is stored for each history object, the above method can not be used. Therefore, as shown in FIG. 6B, the start time of the period query is denoted by Ts and the end time is defined by Te. Search the web page from the defined variables line by line based on "(comma)", and find the Date in the first yyyy type greater than or equal to Ts. The text is collected until the first date that is greater than or equal to Te is found from the text at the Ts start point satisfying the above conditions, and is regarded as data corresponding to the period.

In addition, this process is performed for each history object, and a method of collecting and storing the history object as one file is used.

The association degree determining unit 230 determines the degree of association between the history objects extracted by the history object extracting unit 220. [ In this case, the degree of association is calculated by applying an Euclidean distance algorithm, and the degree of association is calculated by the following equation (1).

Where p and q are historical objects, and n is the frequency.

The Euclidean distance is generally used as a formula for finding the distance between two points in a multidimensional space. It is assumed that the degree of association between two historical objects is higher as the history object is frequently aired in the text, And determines the degree of association by summing the values obtained by dividing the distance between the individual historical objects included in the history information.

In other words, the closer the distance between the two historical objects is, the closer the sum result of each history object becomes (closer to 0).

For example, if it is assumed that the period query is generated as '1501-1510' and history information about 'person' and 'event' within the period is to be known, as shown in FIG. 7A, Extracts N history objects above the frequency.

Also, the number of the history objects and the frequency of the objects included in the extracted number of frequencies are shown. In this case, the number of the historical objects and the frequency is shown in FIG. 7B, The historical objects can be sorted in order of frequency with respect to the data extracted in the period range of the period query '1501-1510' among the pages corresponding to the pseudonymization.

In addition, if the Euclidean distance is calculated based on the data arranged in FIG. 7B, it is confirmed that the frequency of the historical objects included in the two historical objects is calculated based on the ' .

In this case, since all the history objects included in the calculation between the two historical objects are used, if a history object that does not include one is included in the other, calculation is performed including the history object.

Therefore, the determination of the association between two historical objects can be performed only through a single calculation. As shown in FIG. 7C, the result of the determination can be expressed as' 628.037 ', which is the value of' Likewise, it can be confirmed that '628.037', which is the value of 'light', has the same value.

In this way, it can be concluded that the historical object 'JoKwangJo' has a high degree of association in the order of 'heart, mystery, Kim Jung, Nam-gon and Kim An-ro · '.

The detail relation determining unit 240 determines a detailed relationship between the historical objects extracted through the history object extracting unit 220. The detailed relationship judgment is judged by analyzing the contextual elements of the web page and the old document containing the historical object.

The contextual element is a word that implies a relationship between two people in the process of describing a historical object. For example, "Namgon and others must secretly meet and meet with the middle class, and they must discipline them because they are making disorder and making disorder. The word 'kotan' in the sentence 'Doosan gaebyeok and mysterious poetry' can be considered to contain the hostile relationship between 'Kho Gwangjo' and 'Namgong'.

The details of each historical object can be defined in [Table 1] below, and the contextual elements representing each detail relationship can be defined in [Table 2].

Table 3 is the result of analyzing the contextual elements between 'Yi Sun-kyun' and 'Yi Sung-yoo Sungryong' in the detail relation determination unit 240 to determine a detailed relationship between the historical object 'person-person'.

As shown in [Table 3], the contextual elements of 'Yi-Won Kyun' can be seen to be found about 3 times more negative elements than positive elements, The relationship can be defined. On the other hand, the contextual elements between 'Yi Sun-yoo' and 'Yoo Sung-ryong' can be defined as a positive relationship between the two, as positive elements are found to be found more than five times as much as negative elements.

The history information providing unit 300 generates an input query to the history information management unit 100 and provides structured history object information in the history information structure unit 200. That is, the history information providing unit 300 is an interface part for intuitively providing the user with historical information that is searched and clustered corresponding to the user's query in the time zone searching and clustering part. The retrieval result of the historical information is expressed in the graph form between the historical objects clustered by the association rather than the list form of the web document according to the ranking. Here, the search result can be dynamically changed from a search result to a new search result through a roll-in back operation of the user and a menu selection on the interface, thereby reducing the information filtering process of the user. The vertex of the graph corresponds to each history object and is connected by an edge depending on whether there is an association or not. Accordingly, the user can directly or indirectly grasp the atmosphere or the person relationship in the history period by the relationship between the clustering historical objects instead of the process of opening the document and filtering the information from the search results in the existing search engine, Learning can be performed directly without a new search. In addition, it is possible to obtain information on the time-variant of historical information by generating a period query for successive time periods and comparing the respective search results.

The history information providing unit 300 includes a search input unit 310, a query generating unit 320, and a history information display unit 330.

The search input unit 310 serves to input an input query of history information to be searched. 8, the search input unit 310 includes a search type input unit 311 for setting a period-based or object-based search type, an object setting unit 313 for setting a history object to be searched, A keyword input unit 315 for inputting a specific keyword of the period query, and a period query input unit 317 for selecting and inputting a specific period query by listing the period queries.

Here, it is possible to obtain various search results by only selecting the period when the user does not know the exact keyword for the history object or can not specify the keyword with the keyword through the search input unit 310, have.

The query generator 320 generates an input query input from the search input unit 310 and transmits the input query to the history information management unit 100.

The history information display unit 330 displays the result information retrieved from the history information structuring unit 200. The history information display unit 330 displays a relationship graph and a detail relationship between the searched history objects in a relationship graph. The vertices of the relationship graph correspond to each of the history objects, and are connected to each other depending on the presence or absence of association.

In addition, it is preferable that the relationship graph is expressed in different colors according to the detailed relationship between the historical objects determined by the detailed relationship determination unit 240. [

9A to 9C are diagrams showing an embodiment of using the history information providing unit 300. FIG. 9A is a diagram showing a history information providing unit 300 using the period query input unit 317, , A graph of relationship graphs can be provided that allows the user to select a specific period and recognize historical object relations of the period at a glance.

In addition, the history information display unit 330 according to the present invention may further include an appearance frequency graph 333 indicating the frequency of occurrence of the retrieved history object in each period query.

9B is a diagram showing a result of searching for a history object 'person' through the object setting unit 313 and inputting 'light modulation zone' through the keyword input unit 315, When the '1510' age in which the light modulation apparatus frequently appears is generated through the appearance frequency graph 333 displayed on the input unit 317 and a query is generated, it is possible to search for the corresponding query.

9C is a diagram showing a result obtained by selecting 'person' and 'event' as history objects through the object setting unit 313, inputting 'light modulation zone' through the keyword input unit 315, , The search result combining the keywords of 'Kwang-Jo Kwon' with the period query can be provided for the historical objects of 'person' and 'event'.

9D is a search result obtained by inputting the period query '1570-1580' and the period query '1620-1620' using the period query input unit 317. As a user selects a specific two periods, A graph of the relation between history objects and search results about generation and disappearance of historical objects in each corresponding period can be provided through the unit 340.

As described above, according to the present invention, a user can acquire history information and related information by generating a period query only by selecting a period without specifying or inputting a keyword for searching, and the user can intuitively By providing a structured graph that can be grasped, the information related to the history object can be grasped at a glance.

In addition, it is possible to develop a search service applying a change over a period based on a conventional history search service of a search engine, and also build a database that determines the degree of association of related keywords for each keyword, It is possible to provide a search service of new historical information.

Although the present invention has been described in detail with reference to the above embodiments, it is needless to say that the present invention is not limited to the above-described embodiments, and various modifications may be made without departing from the spirit of the present invention.

100: History Information Management Department
110: Data collecting unit
120: Data Study
130: Data server unit
200: History Information Structuring Department
220: history object extracting unit
230: Associated Island Government
240: Detailed relationship judgment section
300: History information service
310: Search input
320: query generator
330: History information generator

Claims (17)

A history information management unit for collecting historical information through a search engine and processing and storing collected historical information based on a period query for searching for information contents based on a change in time;
A history information structuring unit for extracting and clustering relationships between historical objects retrieved from the history information management unit according to the period query,
And a history information providing unit for generating an input query in the history information management unit and providing structured history object information in the history information structuring unit,
The history information management unit includes a data collection unit for collecting web page information and old document information including a history object from a search engine; A data processing unit processing the history information and the old document information collected by the data collecting unit according to periods and sorting the frequency of the history objects; And a data server unit for converting the history object and the annotated information processed by the data processing unit into a time index database.
The history information retrieval system according to claim 1, wherein the history information management unit databaseizes the processed history objects and the annotation information in a time index.
delete The method of claim 2,
The historical object is defined as a person who refers to an actual person described in the history, a cultural property which is a valuable or intangible asset left from the past, an event that is the beginning or center of history development, Wherein the history information retrieval system includes at least one of the history information retrieval system and the history information retrieval system.
The method according to claim 1,
Wherein the data collecting unit includes a crawler for collecting keywords of the history object, collecting web page information including the keyword, and a wrapper for collecting the old document information. Search system.
The method according to claim 1,
The data processing unit may include a morpheme analysis unit for analyzing morphemes of the insoluble tag removing unit, the web page information and the old document information, which removes the insoluble tag of the web page information and the old document information collected by the data collecting unit, A history object sorting unit for sorting the history information of the web page information and the old document information according to the frequency of the history object, Information retrieval system.
The method according to claim 1,
The data server unit includes a web page storage unit for storing the web page for each historical object, an old document storage unit for storing the old document information, a frequency sequential document storage unit for storing sorted old document information in the order of frequency of the historical objects, A history web page storage unit for storing sorted web pages in order of frequency of history objects, and a net web page storage unit for storing sorted web pages in order of frequency of history objects.
The history information structuring unit according to claim 1, wherein the history information structuring unit comprises: a history object extracting unit for extracting a history object of the web page and the old document information retrieved from the history information managing unit;
An association degree determining section for determining a degree of association between the extracted historical objects,
And a detailed relation determining unit for determining a detailed relationship between the extracted historical objects.
The system of claim 1, wherein the input query includes a period query that defines a period of searching.
The method of claim 9,
Wherein the input query further comprises a keyword query that specifies a keyword of the history object to be searched.
The method of claim 8,
Wherein the association degree determination of the associative mapping board is performed by applying an Euclidean distance algorithm, and the degree of association is calculated by the following equation.

Where p and q are historical objects, and n is the frequency.
The method of claim 8,
Wherein the detailed relationship determination unit of the detailed relationship determination unit analyzes the contextual elements of the web page and the preexisting document information including the historical object to determine the historical information.
The method of claim 9,
Wherein the history information providing unit comprises: a search input unit for inputting an input query of a target to be searched;
A query generating unit for generating an input query input from the search input unit and transmitting the generated input query to the history information managing unit;
And a history information display unit for displaying the result information retrieved by the history information structuring unit.
14. The method of claim 13,
The search input unit may include a search type input unit for setting a search type based on a period or an object, an object setting unit for setting a history object of a search target, a keyword input unit for inputting a specific keyword of a search target, And a period query input unit for selecting and inputting a query.
14. The method of claim 13,
Wherein the history information display unit includes a relationship graph that displays a degree of association and a detailed relationship among the searched history objects.
16. The method of claim 15,
Wherein the history information display unit further comprises an appearance frequency graph that indicates the frequency of occurrence of the searched history object in each period query.
14. The method of claim 13,
Wherein the history information providing unit provides a history object that has been changed in each period so as to select any two periods and compare the retrieved results, and provides a relationship diagram graph showing a relationship degree and a detail relationship. Historical information retrieval system.

Images