Jarg Corporation Seeks Sponsors Who: Identify Solutions To Problems With Our Pilot Demonstrations of: Effective Semantic Use of large Ontologies (UMLS) Effective Achievement of Both Excellent Semantic Precision & Recall of Results Effective High-scale & High Performance (Google-like) Database Architecture Show Operational Life Science Applications - as Future Potential for: FEA Search & Retrieval Extraction From New Docs, “concepts” to match / append Agency Official Meta Tags Ideas for Semantic Domain Collaborations in: Avian Flu Pandemic threats Early detection of developing disaster recovery problems Early detection of developing terrorist threats December 6th Collaboration Workshop
Core Base Search & Retrieval Interoperability, Beyond METADATA Semantic Query Access and SW Agent Alerts Dynamic Situation-Awareness, SW Agent-based Search Unified “ Content Awareness ” Throughout The Federal Enterprise Multimedia ’ s “ Native Content ” & Geospatial Search Jarg – SemanTx Life Science Unique-Identifier Combined Index Ontologies
Your Well-Articulated Need ! Domain Ontology’s Contextual Meaning Cluster Pattern Match Syntactic Taxonomy Entity Extraction Word Match/Key Words Directory Ranked By Fit-To Context Bottom-Up Filtering Clearforest Quigo (categorization) Google MSN Verity, Convera, Endeca iPhrase Yahoo Inxight, Fast Autonomy Search Today Top-Down Semantic Results
Filtering By Meaning A Fresh, Scalable, High Performance Approach Sub-Ontology based semantic object-parsing –Enables capture of context for the extraction of “understood features” from within all forms of information to be “semantically represented” then indexed in a common semantic (“fragment”) unique-identifier format Semantically-rich (complex queries) express the context of your need –Return a “collage” of rich-media results –Each result prioritized by its contextual fit to a user’s need
Knowledge Representation Extracted development test produces is_a measures eukaryotic telomericeukaryote recombinationtelomerecell chromosome process Is_a Is_a property_of location_of Telomeres, the physical ends of chromosomes, are essential for maintaining chromosome stability and structure. The mechanisms that maintain the simple sequences present at the telomere within a discrete distribution is poorly understood. One such mechanisms, termed rapid deletion events (RPD) has been described in our laboratory to occur frequently in Saccha- Development of an Assay for Eukaryotic Telomeric Recombination assay Both The Info Source and The Query Ontology’s Query Expansion
Semantic Index NLP Extractor Index Server Natural Language The Treatment of Diabetes The slifo soifgo isnof odsnofg no snfnskjnfkjsnkfjnnskjnfkjfnksdj sdkjvcnksjnskjfnksjndkj slfkm gjdg dlkf dlsdlkg diufiv dfn glkdd dgfrflgf dlfgjedsd flirflfglfkd ;elk fgdissolf ;gldisj slif fglfgikd ldf fli j ldjvg;ods fid sdlfv fvfdgh kujgijr ibligldkngb lkgoidlfiglirfg bfirifgi kdfjglfif fldkfgljf disdfhrfog rpiotg fjioeijr fgoijed oifj drdgfoijfg oidj doifoifj dijdoidfj dij ddoif fjfoidjdoj Document Doc Parser Doc Parser Ontology NLP Extractor Index Server DocParser - translates formats to ASCII for extraction NLP – syntactic analysis Extractor - uses ontology to extract key concepts and relationships Ontology - domain-specific concepts and relationships Index Generator - creates Semantic Index Index Servers - identify documents that match user queries Semantic Index – index of ideas How Jarg’s Platform Works Go Search Results Asljeo fof oe ojf erij rriro oijoi E er pepdoj boigboie oio Qoijd owfj wojf oj oqi aoc a au Asljeo fof oe ojf erij rriro oijoi E er pepdoj boigboie oio Qoijd owfj wojf oj Asljeo fof oe ojf erij rriro oijoi E er pepdoj boigboie oio Qoijd owfj Query Natural Language Text Keynets (xml) Query Fragments Semantic Index Document Fragments Semantic Index
Step 3: Review highlighted contextual answer within document Step 1: Enter query in plain English Step 2: Proves match results Process Overview & Advantages Answers returned, ranked by contextual relevance (solves relevance and scalability issues) Allows for cross-disciplinary research to shorten drug discovery & clinical response cycles
Understands ? “Graph” Why !
“Articulate” your need with lots of Context
Click to MedLine
Mass General Lab
Many Joins
Mass General Missing Only 2 Joins
Term Synonym Found Mass General Lost the others
Semantic Federal Doc & Meta-Tagging Check-off Suggestions ?
Charts & Graphs Extension Vertical Mkt: Life Sciences Knowledge Engine Limited Index Release Grouping with Visual Graphics Distributed Index Release Structured Data Extension Intelligent Agents Images & Video Extension Re-use & Mediation of Ontologies Queries & Cluster Analysis Sound & Music Extension Summarization & Comparisons Data Warehousing & Mining Semantic Knowledge Indexing Platform Current (SKIP) Rollout Plan High Perfor- mance Indexing Engine Text SDE GraphImagesSound Available to beta accounts IP Partnering Opportunities
Core Base Search & Retrieval Interoperability, Beyond METADATA Semantic Query Access and SW Agent Alerts Dynamic Situation-Awareness, SW Agent-based Search Unified “ Content Awareness ” Throughout The Federal Enterprise Multimedia ’ s “ Native Content ” & Geospatial Search Jarg – SemanTx Life Science Unique-Identifier Combined Index Ontologies
Jarg Corporation/SemanTx Life Sciences 330 Bear Hill Road, Suite 230 Waltham, MA (USA) Attn: Michael P. Belanger, x206, Thank you for inviting us today. Links to resources: “Semantic PubMed”