Semantic Search Jiawei Rong 2005.11.30. Authors Semantic Search, in Proc. Of WWW 2003. Author R. Guhua (IBM) Rob McCool (Stanford University) Eric Miller.

Semantic Search

Jiawei Rong

2005.11.30

Authors

Semantic Search, in Proc. Of WWW 2003. Author

R. Guhua (IBM) Rob McCool (Stanford University) Eric Miller (www.w3.org)

Semantic Web Definition

Semantic Web is the idea of having data on the web defined and linked in a way that it can be used for more effective discovery, automation, integration , and reuse across various applications.ResourceRelation

Semantic Web Model

Directed labeled graphNodes: ResourcesArcs: Property Type

Semantic Web vs. Current Web

Documents vs. Real World Objects

Human vs. Machine Readable Info

HTML vs. Semantic Web

Distributed Extensibility

Semantic Search Introduction

Navigational Search--The user provides the search engine a

phrase or combination of words which are expected to be found in the documents. No interpretation.

Research Search--The user provides a phrase which is

intended to denote an object about which the user is trying to gather information.

Semantic Search Goals

Augment traditional search results with data pulled from the Semantic Web

Use an understanding of the denotation of the search term to improve traditional search

TAP

TAP (Stanford University)An infrastructure for the applications on the

Semantic WebFor sites to publish data onto Semantic WebFor applications to consume the data throug

h query interface called GetData

GetData query interface

GetData(<resource>, <property>)=><value> Examples

GetData(<Yo-Yoma>, birthplace)=><Paris, France> GetData(<Paris, France>, temperature)=>57F GetData(<EricMiller>, livesIn)=><Dublin, Ohio>

Search: returns all the resources whose properties contain the query string

Reflection: Given a node, returns the arcs going in and out of the node

TAP Publish, Registry and Cach

TAPache Creates a directory and places RDF files in it. The

URL associated with each graph is that of the file.

Registry A separate server to keep track of what graph has

the values for which properties about which resources.

Caching Cache the responses to GetData Requests

Data Source

Activity Based Search (ABS)Musicians, athletes, actors, places, products

W3C Semantic Search People, W3C activities, working groups and

other committees, documents and news

System Architecture

Choose a denotationMap the search term to one or more nodes

of the Semantic WebDetermine what to show

What data to show and in what orderFormat

Display the results in a user interface

Choosing a Denotation

Ambiguity SolutionPick the denotation according to the

popularity of the termOr according to the user profileOr according to the search context

Complex Search Term SolutionRestrict complex search terms to only two

denotations

Determining what to show

Method 1 Breadth first order search with heuristics rules such

as include at most N triples with the same source and same arc label or include at most M triples with the same source.

Method 2 For each class or object, the set of pre-defined

properties are manually specified

Hybrid

Formatting

Semantics for Text Search

Enable a search engine to understand that different occurrences of the same string denote different things

Filter and rank the results to show the documents referring to the chosen denotation

Heuristics

Knowing the user is searching for a person

The type of the personGeneral knowledge about the person,

like e-mail address

Conclusion

Two mechanismsAugment the search resultsProvide denotation to the search

Future WorkHelp the text search system to exploit a

deeper understanding of the search term’s denotation

Thank you!