TMS

From Saltlux

Concept of [IN2]TMS
• As an information mining solution that accomplishes information retrieval, auto summarization and auto classification, [IN2]TMS significantly reduces time for search, analysis and application of knowledge information.
• [IN2]TMS is a system developed to discover the knowledge hidden in huge piles of documents and to organize these knowledge information. Through analysis algorithm based on huge pile of language resources, NLP technology and rules and statistics, it provides best functions customer needs.

Key Characteristics of [IN2]TMS


1. Qualitative Information Extraction
• Accurate multilingual analysis: Text mining system is largely depended on NLP technology. [IN2]TMS is based on the world class multilingual analysis technology.
• Statistics and rule based Hybrid information retrieval engine: Information retrieval system can apply machine learned statistical algorithm and uses XRE pattern matching engine that is possible to be functioning strong and soft pattern rule as hybrid.
• Machine Learning: For statistic based information retrieval, quality of statistical prior learning is very important. And [IN2]TMS provides automatic machine learning tool for text mining.
• Expansion to core search function: Provides core document search function by retrieval and indexing of core concept or keyword of documents.

2. Automatic Document Classification
• Diversified classification model: [IN2]TMS supplies various classification algorithm including Rochio, Naïve Bayesian, SVM for classification.
• Auto classification learning function: Prior learning machine is built-in according to classification system for statistic based automatic document classification and the user customized classification function is provided.
• Multilevel document classification and management tool: [IN2]TMS supports multilevel classification learning and auto classification system functions according to user designed taxonomy classification system.
• Excellent auto classification performance: Auto classification tool has been realized through integration of rule based auto classification technology to SVM classificationengine that shows the best performance among statistical methods.

3. Automatic Document Summarization
• Keyword and retrieval summarization: Core keywords, concepts and sentences are extracted, and summarization function is provided based on these factors.
• Template base creation summarization: [IN2]TMS provides summary creation function base on information retrieval rules through selection of summarization template by each classification through automatic classification of summary target documents.
• Multilingual auto summarization: Auto summarization system of [IN2]TMS shows high performance against various languages including English and Japanese in addition to Korean.

4. Automatic Document Clustering
• [IN2]TMS enhances information accessibility through clustering of documents with similarity and relativity.
• Attribute retrieval and similarity calculation: Similarity and relativity of attributes are calculated based on genetic attributes of documents retrieval, vector built according to types and importance.
• Similarity calculation model: [IN2]TMS enables to selectively use of Cosine, Euclidian, Jaccard, Dice similarity measurement for clustering based on k-means.
• Multilingual document clustering: Clustering system of [IN2]TMS can make clusters on the multilingual documents through application of ontology and transformation and expansion of multilingual attribute vectors.
• Example based search system: As a typical application solution, query by example of document clustering may be applied to various parts.

Systematic Structure of [IN2]TMS


Platform
- Intel x86 : MS Windows 2000/2003Server, Linux
- SUN SPARC : Solaris 8(SUN OS 5.8)
- HP PA-RISC: HP-UX 11.x

Best Practices
KIPO (December 2004년) – Previous technology automatic comparison system
LG Electronics (September 2006)- Intelligent previous technology search system

History
Grand Prix for New Software Award in August 2003
GS (Good Software) Certification in September 2007