50 likes | 171 Vues
This document details processes and tools for converting XML schemas and DTDs to DAML format, emphasizing the challenges in capturing complex XML elements and attributes. It introduces the Query Relevance Assessor (QRA), demonstrating how DAML improves text retrieval accuracy through relevance ranking, leveraging semantic information from DAML ontologies. The document highlights significant achievements, including the development of reusable Java components for DAML transformations, while addressing current challenges like scalability and the need for partial DAML conversion.
E N D
BAH DAML Tools • XML To DAML • Query Relevance Assessor • DAML XSLT Adapter
XML To DAML • The high level tasks in order are • XML Schema to DAML - current task • XML DTD to DAML - via XML Spy as a preprocessor • XML Content to DAML - dependent on the language map • XML Schema to DAML Accomplishments • Import problem solved: Graphs of XML schemas can be translated over time • Core XML Schema components currently supported • Namespace & nonamespace schemas • Simple/complex element declarations & Attribute declarations • Global/anonomous attribute groups, model groups, & complex type declarations • sequence (currently without order semantics), choice (based on a daml set difference), & all • attributes & attribute refs, simple/complex elements & element refs, & group refs • The language map - cascading Restrictions model nested XML schema components • Challenges • The harder portions of the language map • includes • fixed/default/null values, mixed/empty content, derivation by restriction/extension • order semantics, elemement substitution groups, abstract elements/types, block/final, wildcards
Query Relevance Assessor(QRA) • Query Relevance Assessment with the QRA is HOT. . . • Demonstrates how DAML can be used to enhance the accuracy and precision of WWW text retrieval by relevance ranking DAML mark-up • Exploits the wealth of contextual knowledge contained within DAML ontology classes, properties, and instances • Imparts a semantic component to relevance rankings of DAML-annotated text with respect to query terms when only minimal query context may be available • Achieves a minimal loss of semantic information important for relevance ranking • . . . yet significant CHALLENGES remain • Incorporate indirect relevance ranking and free text search • Address semantic heterogeneity • Support multi-term queries in a semantically appropriate fashion • Investigate incorporation of machine learning methods for categorizing DAML mark-up
DAML XSLT • What’s Hot with DAML XSLT! • Enables an end user to perform a myriad number of unique transformations on a DAML resource (mark up) leveraging current XSLT technology. • Makes DAML useful to a much wider audience of users (it offers a quick and easy way to exploit the power of DAML, and package it into a familiar form, such as: XML, HTML, WML, BizTalk, etc.) • Accomplishments • Built a “new” DAML Instance Model and converted DAML to representative XML by encoding DAML cycles into “cycles-less” XML. • Created a powerful, reusable Java component for DAML transformation • Challenges • Scalability - Large Data Sets -How much is too much? • Currently No “Partial” DAML Conversion • Add Input Filters to pare down large data sets • Choice of “Root Node” for the Representative XML • Ordering of XML Elements • Currently uses alphabetical sorting of Subjects and Predicates • Use a “style sheet” to impose ordering/filtering/sorting of elements • Continued Development of a robust DAML Instance Model • Focus on instance data creation, visualization, navigation, and manipulation • Multi-Pass DAML Translation • Current XSLT is single-pass only • Leverage DAML cycles, relationships, etc.
BAH DAML Website • The July PI meeting presentations can be found at http://www.davincinetbook.com:8080