swlee: analytics for noisy data in ontology

- error type (from parsing error, triplet error, ontology mapping error): debugging purpose - error detection (by manual or by some analysis): First we must analyze the results by human.
1. parsing structure fault (disconnected graph)
2. triplet - mismatch between phrases and result
3. ontology mapping - WN, FrameNet, SUMO

- error - automatic correction
1. if POS ambiguity, change POS for some error around the mal-functioned component in the parsing structure
2. triplet - parsing error - failure of deteting the phrase - error correction
3. ontology mapping -- error detection - error correction

- error processing UI

- performance analysis

to be submitted for AND (due 13/April)

---
KEOD
- to integrate and analyze the ontology engineering projects so far
- swlee: how to enhance SIET and merge CoreontoWeb - to incorporate relation classification
- dhchoi: how to make a midterm results for summarization by information entropy and ontology
- jhahn: how to contribute isa hierarchy detection algorithm and how to apply 2008-results
- kek: evaluate them objectively and suggest a roadmap for pratical purpose or some possible goal

---
For Korean problem,
1. prepare Korean-English corresponding Wikipedia page
2. try to use KAIST-made QA Trec-like testsuite

--
MT Summit
1. English-Korean comparing paper - ontology development synergetic approach
- synergetic effect
- translation between two languages of ontology, transliteration etc.

--
how to make our website SemanticWeb
rf. 1. http://www.mindswap.org/, 2. http://www.saltlux.com/

application of UIMA - Korean part generation
- 형태소해석기 - C2Java, Perl2Java - 문제파악 --> 맡긴다.

---

=====