Philosophy Learning for Chinese Data Association and Information Disclosure in Ethnology and Human studies.


75 views
Uploaded on:
Category: Art / Culture
Description
Chinese Academy of Social Sciences. twentieth CODATA International Conference Beijing, 23-25 ... Cosmology learning casing for data association and information revelation ...
Transcripts
Slide 1

Cosmology Learning for Chinese Information Organization and Knowledge Discovery in Ethnology and Anthropology Kong Jing Institute of Ethnology & Anthropology, Chinese Academy of Social Sciences

Slide 2

Outline Introduction Definition of Ontology learning Development of Ontology taking in Our examination target Ontology learning outline for data association and information revelation CHOL(a Chinese Ontology Learning Tool) Architecture Components Approaches Experiment in Ethnology and Anthropology Conclusion & Future Work twentieth CODATA International Conference Beijing, 23-25 October 2006

Slide 3

Definition Ontology learning is characterized as the arrangement of strategies and systems utilized for building a philosophy sans preparation, enhancing, or adjusting a current metaphysics in a self-loader design utilizing a few sources. (A. Gómez-Pérez, D. Manzano-Macho. A study of cosmology learning strategies and Techniques. OntoWeb Deliverable D1.5, 2003,6) twentieth CODATA International Conference Beijing, 23-25 October 2006

Slide 4

Development Recently, there has been a surge of enthusiasm for concentrating on philosophy learning. In 2000, the main workshop on metaphysics learning held in conjunction with the fourteenth European Conference on Artificial Intelligence (ECAI2000). In the previous years, numerous metaphysics learning apparatuses, for example, TextToOnto 、 OntoLearn 、 OntoLT 、 Adaptiva 、 the ASIUM framework 、 the Mo\'k Workbench 、 SOAT and DOGMA have been produced. twentieth CODATA International Conference Beijing, 23-25 October 2006

Slide 5

Our exploration objective Despite the noteworthy measure of work done on metaphysics learning as of late, taking in cosmology from Chinese content hasn\'t been broadly connected by and by. So our exploration target is to ponder the utilization of cosmology learning in Chinese data association and information revelation. twentieth CODATA International Conference Beijing, 23-25 October 2006

Slide 6

Ontology learning outline for data association and information disclosure

Slide 7

CHOL (a Chinese Ontology Learning Tool) Architecture Components Approaches

Slide 8

CHOL Architrchture

Slide 9

CHOL Main Modules Initial Ontologies Components of CHOL twentieth CODATA International Conference Beijing, 23-25 October 2006

Slide 10

CHOL Main Modules Text Processing Extraction of Candidate Term Identification of Domain Term Extraction of Relations Formal Representing twentieth CODATA International Conference Beijing, 23-25 October 2006

Slide 11

Initial Ontologies CNLO Chinese Natural Language Ontology incorporates all the fundamental Chinese lexical words and the lexical relations between the Chinese-dialect ideas. It " s utilized for content preparing and lower-level ontologies removing. It contains lexical information of Chinese. Top-Level Ontology CGDO Chinese Global Domain Ontology Second-Level Ontology Chinese Foundation Domain Ontologies CFDO 1 , CFDO 2 , CFDO 3 , … Third-Level Ontology CSDO 1 , CSDO 2 , CSDO 3 , … Chinese Specific Domain Ontologies Bottom-Level Ontology twentieth CODATA International Conference Beijing, 23-25 October 2006

Slide 12

Initial Ontologies CNLO Chinese Natural Language Ontology incorporates ideas of all particular area and taxonomic relations between ideas. It " s utilized for learning Completeness and lower-level ontologies extricating. Top-Level Ontology CGDO Chinese Global Domain Ontology Second-Level Ontology Chinese Foundation Domain Ontologies CFDO 1 , CFDO 2 , CFDO 3 , … Third-Level Ontology CSDO 1 , CSDO 2 , CSDO 3 , … Chinese Specific Domain Ontologies Bottom-Level Ontology twentieth CODATA International Conference Beijing, 23-25 October 2006

Slide 13

Initial Ontologies CNLO Chinese Natural Language Ontology Top-Level Ontology for every particular space its establishment metaphysics is developed. Every particular area has some foundational areas. Its establishment metaphysics incorporates ideas of its foundational spaces. CGDO Chinese Global Domain Ontology Second-Level Ontology Chinese Foundation Domain Ontologies CFDO 1 , CFDO 2 , CFDO 3 , … Third-Level Ontology CSDO 1 , CSDO 2 , CSDO 3 , … Chinese Specific Domain Ontologies Bottom-Level Ontology twentieth CODATA International Conference Beijing, 23-25 October 2006

Slide 14

Initial Ontologies CNLO Chinese Natural Language Ontology Top-Level Ontology incorporates ideas of one particular area. It gives nitty gritty portrayal of the area ideas from a confined space. CGDO Chinese Global Domain Ontology Second-Level Ontology Chinese Foundation Domain Ontologies CFDO 1 , CFDO 2 , CFDO 3 , … Third-Level Ontology CSDO 1 , CSDO 2 , CSDO 3 , … Chinese Specific Domain Ontologies Bottom-Level Ontology twentieth CODATA International Conference Beijing, 23-25 October 2006

Slide 15

Our methodologies Initial ontologies Constructing CNLO CGDO CFGO CSDO Concepts extraction Method Relations extraction Algorithm twentieth CODATA International Conference Beijing, 23-25 October 2006

Slide 16

CNLO Constructing Mapping Hownet into Natural Language Ontology. Results Chinese lexical ideas: 68,273 Relations Synonym: 60,310 Act/result : 7,121 twentieth CODATA International Conference Beijing, 23-25 October 2006

Slide 17

CGDO Constructing Mapping Chinese Classification Thesaurus into Global Domain Ontology Results Chinese Term: 115142 Concepts: 128747 Relations: Synonym: 19158 Generality: 41714 Hierarchy: 67830 twentieth CODATA International Conference Beijing, 23-25 October 2006

Slide 18

CFGO & CSGO Constructing CFGO Constructing Each CFGO of CSDO is progressively built from CGDO by selecting the ideas of it\'s foundational areas. CSDO Constructing The underlying CSDO is developed from CGDO by selecting the ideas of every area. Utilizing cosmology learning technique, the underlying CSDO will be self-loader redesigned and enhanced by CHOL. twentieth CODATA International Conference Beijing, 23-25 October 2006

Slide 19

Concepts extraction Method Domain term distinguishing proof recipe For every competitor term the accompanying term weight is figured: DR t,k measures the area importance of a term t in a space D k . DC t,k measures the dispersed utilization of a term t in a space D k . GC t measures the appropriated utilization of a term t in all spaces. twentieth CODATA International Conference Beijing, 23-25 October 2006

Slide 20

Relations extraction Algorithm Input: another found term t & reports in which this term is utilized. Yield: Relations between term t and related terms Step1: Extract all terms in CGDO and new terms found by CHOL from archives. Every archive is communicated as a weighted catchphrase vector comprised of all terms for SOM calculation. Step2: Use SOM for term grouping and create bunches of term. Step3: Use the fluffy bunching calculation to create the two level progressive system relations of terms. Step4: Use our area term recognizable proof strategy to distinguish the areas to which term t have a place. On the off chance that term t have a place with various area, for every space creates a term relations tree. Step5: Trim and overhaul these term relations trees utilizing CGDO and CNLO. twentieth CODATA International Conference Beijing, 23-25 October 2006

Slide 21

Screenshot of CHOL twentieth CODATA International Conference Beijing, 23-25 October 2006

Slide 22

Experiment in Ethnology and Anthropology We have tried CHOL in ethnology and human sciences to discover and remove obscure term and the relations between terms from Chinese content about minority custom in China. twentieth CODATA International Conference Beijing, 23-25 October 2006

Slide 23

Example: CHOL connected in Chinese minority celebration database. Separated ideas: " 雪顿节 (Xuedunjie)" 、"望果节 (Wangguojie)" 、"法会 (Fahui)" 、"三月街 (Sanyuejie)" 、"采花山 (Caihuasan)" 、"姊妹节 (Zimeijie)"… … Extracted relations: " 瑶族 (Yao)"- " 盘王节 (Panwangjie)" " 畲族 (She)"- " 乌饭 (Wufan)" " 藏族 (Tibetan)"- " 转山会 (Zhuanshanhui)" … twentieth CODATA International Conference Beijing, 23-25 October 2006

Slide 24

Precision and review for the phrasing ID twentieth CODATA International Conference Beijing, 23-25 October 2006

Slide 25

Conclusion & Future Work We have built up a model framework for philosophy gaining from Chinese corpus, named CHOL. In CHOL, we propose a few techniques to recognize term of area and to remove taxonomic relations between terms. These strategies are turned out to be attainable and successful in utilization of data association and learning revelation in ethnology and human sciences. At present, CHOL is only a basic model framework. In future, we will utilize more strategies, particularly, profound semantic examination. CHOL will be connected in more distinctive space and bigger datasets. twentieth CODATA International Conference Beijing, 23-25 October 2006

Slide 26

Thanks kongjing@cass.org.cn

Recommended
View more...