Untangle A New Ontology For Card Catalog Systems

Untangle a new ontology for card catalog systems

File Name: AAAI00-205.pdf

File Size: 424.77 KB

File Type: Application/pdf

Last Modified: 17 years

Status: Available

Last checked: 29 days ago!

This Document Has Been Certified by a Professional

100% customizable

Language: English

We recommend downloading this file onto your computer


From: AAAI-00 Proceedings. Copyright © 2000, AAAI (www.aaai.org). All rights reserved

Untangle: A New Ontology for Card Catalog Systems
Christopher Welty and Jessica Jenkins
Vassar College Computer Science Dept

Poughkeepsie, NY 12604-0462
[email protected]
The ontology used by most card catalog and bibliographic Untangle-thing
systems is based on a now outdated assumption that users of
the systems would be looking for books on shelves, and
Event/Object Modality
therefore only books were first-class objects, with people, Document
organizations, etc. as simple attributes. This limited the abil- Audio Visual
ity of a user to browse. A new ontology for card catalog sys-
tems is proposed that suggests that persons, organizations, Book Article
conferences, etc., should be first-class objects with attributes Conference Graphics Text
Object Collection
and relations of their own, creating a rich space of back-
ground information that helps users find what they are look-
Proceedings Journal- HTML PDF
ing for. This new ontology has been implemented in a Person
knowledge-based system called Untangle, which demon- issue
strates two key advantages of this rich information space: it Organization
enables automatic augmentation of the data through reason-
ing, and it enables a new paradigm for search that combines Figure 1: Taxonomy of Untangle concepts

querying and browsing

tion discovered during browsing may suggest new queries,
since clearly users may not have perfect knowledge of what
Introduction they are looking for

We present here the ontology of the Untangle system,
Library card-catalog systems are quite old, with evidence of
and then describe the system itself, which is available on
recorded indexes dating back at least as far as the 12th cen-
the web

tury. Through the centuries the underlying assumption of
these systems – implicit or explicit – was simple: the only
object of any search is a book. This assumption led to an The Ontology
ontology that still exists today: the only type of object in a
The Untangle Ontology contains two types of objects:
card-catalog system is a book, which has attributes such as
things related to topics (TOPICAL-THING) and things
author, title, publisher, date, etc. While we understand that
related to entities in the real world (UNTANGLE-THING)

these attributes correspond to people, organizations, etc., in
A more complete description of topic related issues can be
implemented systems they are nothing more than strings

found in (Welty and Jenkins, 1999), and will not be dis-
As the size of, and access to, libraries increase dramati-
cussed here

cally with modern communication, this ontology no longer
serves, and a new ontology, that supports a new paradigm in
library search, is called for. The new ontology supports the Basic Types
notion that persons, places, events, organizations, etc., are The domain of real entities for Untangle is split into three
first-class objects, and that a user browsing through the main types: Modalities, Documents, and Event/Objects
information may submit individual queries whose results (Welty, 1996). Some of the taxonomy is shown in Figure 1,
may be these objects, not books. For the web, the conse- and in the demo you can select “list the main objects in the
quence of this is the same, the result of an individual query ontology” to get a textual view of the ontology taxonomy

does not always have to be a web page. A search may be for Documents are the central objects in the system. The
information about a person, place, organization, etc. background information is there mainly to support finding
This implies that search must be, in this age of informa- these objects. Although the result of any particular query
tion, an iterative process that alternates between browsing may be any type of object, it is assumed that the result of
the information space, and querying it. The results of que- any sustained interaction will be a document. The docu-
ries will be new starting places for browsing, and informa- ment hierarchy is fairly straightforward, and was derived
from Bibtex with a few extras added for web support

Copyright © 2000, American Association for Artificial Intelligence Modalities are an object type which are used to represent
(www.aaai.org). All rights reserved. the physical manifestations of a document in the real world
(Welty, 1998). A single document, e.g. a book, may have a
published hard-copy version, an on-line PDF version, an paper published in the conference proceedings

on-line HTML version, and an audio-cassette version. Each • A person is interested in a topic if they write a paper on
of these real things are the same book, they share the author, that subject

title, subject, they contain the same information. They each • If a person is an employee-of a department, they are an
also have attributes of their own; on-line versions have a employee-of the organization the department is a part of

URL and format, hardcopy books have a shelving code, etc

In addition to making more sense, breaking these attributes
out of the document object makes it possible to represent Demo Tour
the fact that several web pages may be the same thing, and Click on the small blue “home” icon at the bottom of the
therefore returning a single “hit” for a query instead of one screen to get to the untangle home page, or go to the URL:
per web page helps prune and narrow search results. http://untangle.cs.vassar.edu/. The first three links in the list
Event/Objects are the background information that you see are the main links for exploring the demo

enhances the catalog and makes browsing possible. Events Let’s say you’re looking for articles on ontologies for
are things like conferences, workshops, etc., that often information systems. You remember hearing a talk about
result in people getting together exchanging information this at a conference in Florida some years ago

and publishing papers. It is not unusual, for example, for a
person to search for a paper they remember hearing at a 1. Hit the “query” link on the Untangle home page

conference, without remembering the author or title. 2. Select conference from the taxonomy (under event)

Objects are things like people and organizations. 3. Type “Florida” into the text box labelled “location”, and
click “do it.” The result is a list of conferences in Florida

4. Click on “FLAIRS-96” (you can select it with the radio
Relationships button, but in this case clicking on it is a shortcut). The
Each type of object has a set of attributes and relationships result is a dynamically generated page that describes all
that help define it. To explore the ontology and see the rela- the information the system has about the conference. At
tionships defined for each type of entity, in the demo you this point you may recognize someone in the list of par-
may click on “list the main objects in the ontology” and ticipants. Let’s say you don’t

then click on any concept. Another good starting place is 5. Click on the proceedings. This results in a list of articles
“explore the Untangle topic space” select some topics, and published in the proceedings (and in the knowledge
then look at the entities classified under that topic. Each base). You notice one, “Intelligent Assistance for Navi-
entity has links for all the entities it is related to. gating the Web.” which sounds familiar

The complete set of relationships for each object can be 6. Click on it. You get information on this article. Aha!
accessed through the demo, we provide here the basic set of That’s the one. Now you see in the list of topics “formal
relationships for navigating through the space. ontologies in information systems.”
Articles have an author, and are published-in some kind 7. Click on that topic, and you get a list of people, events,
of COLLECTION, such as a JOURNAL-ISSUE or a pro- publications, and organizations that have been classified
ceedings. Collections may have editors, and are published- under this topic

by an organization. A journal-issues may be an issue-of a
Continue to explore. The icons on the bottom of every page
journal, and a proceedings is the proceedings-of a confer-
go to the home page, to the query page, and to the FAQ
ence. A conference has participants, a location, and is spon-

sored by an organization. A person may have an affiliation,
and may be a student-of some school. We also support
former-employee-of and former-student-of. References
Welty, Chris and Jenkins, Jessica. An Ontology for Subject

Reasoning J. Data and Knowledge Engineering. 31(2):155-182. Sep-
In addition to supporting combined querying and browsing, tember, 1999. Elsevier

the new ontology supports reasoning to enrich the informa- Welty, Chris and Ide, Nancy. Using the Right Tools:
tion space. The main point of using a KR system instead of Enhancing Retrieval of Marked-Up Documents. J. Comput-
a database is that reasoning can be used to enhance the data ers in the Humanities. Summer, 1999. 33(10):59-84. Klu-
that can be automatically mined from existing bibliographic wer

databases. The bulk of the reasoning is used to support the Welty, Chris. Towards an Ontology for Library Modalities

spatial representation of topics used in the system, and to In S. Ali, ed. Proceedings of the AAAI-98 Workshop on Rep-
support data consistency. We describe here a few examples resentations for Multi-Modal Human-Computer Interac-
tion. AAAI Press. July, 1998

of the reasoning over the types described above that are
used to augment the data: Welty, Chris. Intelligent Assistance for Web Navigation

Proceedings of the 1996 Florida AI Research Symposium

• A publisher is any organization that publishes something. May, 1996

• Two people are collaborators if they are authors of the Welty, Chris. A Knowledge-Based Email Distribution Sys-
same document. tem. Proceedings of the 1994 Florida AI Research Sympo-
• A person participates in a conference if they author a sium. Journal of AI Press. May, 1994

Library card-catalog systems are quite old, with evidence of recorded indexes dating back at least as far as the 12th cen-tury. Through the centuries the underlying assumption of card-catalog …

Download Now

Documemt Updated


Popular Download


Frequently Asked Questions

What is catalogers desktop?

Cataloger's Desktop is an integrated, online documentation system with almost 300 of the most important cataloging and metadata resources, including RDA Toolkit, LC policy statements, subject cataloging manuals, MARC formats, and LC Rule Interpretations. List of books on cataloging, with particular reference to RDA.

What does it mean to catalog a book?

To "catalog" a book or other form of library material involves several interrelated processes which all contribute to the achievement of Charles Ammi Cutter's "objects" for a catalog: To enable a person to find a book of which the author, title, or subject is known

What is cataloging and classification?

The fourth edition of the late Lois Mai Chan's classic Cataloging and Classification covers the analysis and representation of methods used in describing, organizing, and providing access to resources made available in or through libraries.

How many standards and guides are needed for cataloging?

Providing complete cataloging for a library collection requires simultaneous use of four, sometimes five, different standards and guides. This LibGuide will explain the tools needed, along with information on the sources for them.