Ontaria Use Cases

Overview

This document is intended to support Ontaria development by clarifying and justifying system functionality and proposed system functionality.

Item costs are given in Ideal Programming Time (IPT), an Extreme Programming notion of how much time it would take if there were no fire drills, meetings, vacations, etc. The releationship between IPT and the calendar depends on the empirical load-factor / project-velocity. Current rate seems to be about 3 IPT days per real week.

A description of actors in these cases is given at the end, although it's a good place to start for brainstorming more use cases.

Search Use Cases

Searching is the first part of selection: identifying the options. Users may want to search for ontologies, ontology components (classes and properties), software, publishers, etc.

Dumb keyword search

This is the barest form of keyword searching, familiar to users from search engines like AltaVista and Google, but with only the simplest form of pattern match, and no ranking of results.

1. User enters one or more words in the "Search" box (which appears on front page as well as most data pages)
2. System responds with list of resources, each briefly described, which have some property with a literal value which contains the word, for each word. Listed resources have links to focus on them.

Response in less than 1 second, typical, when Ontaria hosting 100K triples. Development costs for maintaining response time for searches with more triples are attributed to the harvesting options which produce those triples.

IPT: done

Keyword search expressions

Support for negation (-word), sequencing (word "word word"), and disjunction (word or word) in search expressions.

IPT: 2d

Numeric Sort Options

Sort results by: last-modified-date, number-of-mentions-in-other-data, and number-of-domain-which-mention-it.

IPT: 4d

Strong and weak keywords

Have different kinds of keywords, where stronger keywords cause stronger matches, putting results earlier in search results. For example

1. keywords given explicitly as keywords, by ontology provider or 3rd party
2. words used as rdfs:label or occuring in dc:description, ...
3. taken from all literals (as in dumb keyword search
4. taken from text of URIs
5. inferred from (eg wordnet) links to other keywords. Search for "person" would also find uses of "individual", "mortal", "human", "soul".

IPT: 5d

Only Search for Ontologies

Users can say they only want to see ontologies in results, as opposed to arbitrary resources in ontologies and instance data.

IPT: 1d

RDF (n3) Search Expression

Users can perform arbitrary graph match (with builtins)

IPT: 3d, given n3 parser (2d)

Top Ten Lists

Simple lists on front page of ten most-recently-modified, ten most-referenced, and ten most-referenced-from-different-domains Ontologies.

IPT: 2d

Category Drilldown

Like dmoz (open directory project), Yahoo!, or xml.org (NAICS, UNSPSC).

Would rely on categories provided somewhere in RDF. (See use case [not written] on allowing categorization via Ontaria)

IPT: 8d

Follow Links to "Similar" Resources

Allow surfing links provided in ontology or coming from other data sources, including back-links.

IPT:

Inspection Use Cases

Inspection occurs in the second part of a selection process, as one tries to learn more about each candidate, and in post-selection use as one simply tries to learn how to use the ontology.

View Properties of a Resource

For arbitrary RDF resources, display a table of its properties and inverse properties. Allows RDF instance data, including ontologies, to be views and browsed. URI terms appear as links to Ontaria's view of properties of that resource.

IPT: done

Ontology At-A-Glance

Single page overview of all the elements in the ontology, along with key metadata.

IPT:

Automatic Documentation

Give each element in the ontology a decent web page of documentation based only on the RDF (OWL) ontology file itself. Removes need for extra documention or related web sites. (Similar to Javadoc or Doxygen.)

IPT:

Who Is Using An Ontology?

IPT:

Who Provided an Ontology?

IPT:

What committments have been made

IPT:

Is it OWL Lite, DL, or Full?

IPT:

Is it dereferenceable?

IPT:

Ontaria provides HTML on Ontology-Term URI dereference

User puts class/property/ontology URI into normal browser, gets automatic documentation page. Should work for both Hash and Slash URIs. Requires proper configuration by web host.

IPT:

Monitor Use Cases

Monitoring occurs at several levels, from the feel-good experience of watching the number of triples in Ontaria climb each week to be being notified if a bug is reported in an ontology of interest.

SemWeb Statistics

IPT:

Ontaria Usage Statistics

IPT:

Mail-Me If Something Changes

IPT:

"my ontaria" page of news/updates

IPT:

Publication Use Cases

Publication on the web can be as simple as a mouse click where the user offers a public rating of a software package or as complex as creating an ontology under the auspices of a standards body.

User wants to publish somehow

IPT:

W3C hosts ontology

1. User applies for w3.org namespace to be assigned to ontology, agreeing to Acceptable Use Policy.
2. W3C examines application, appoves it.
3. User uploads ontology
4. Ontaria indexes it automatically
5. User uploads edited ontology as needed
6. Ontaria re-indexes it automatically
7. W3C hosts ontology indefinitely

IPT: 40d

User has Bug/RFE

IPT:

User willing to report on usage

IPT:

Other Use Cases

Cases not fitting into one of the categories above.

Random user finds site, knowing nothing of Semantic Web

IPT:

Ontaria harvests all RDF when used in search, clicked-on

IPT: done

Ontaria harvests ontologies know to daml.org, schemaweb

IPT: 6d to get performance on search back under 1s

Ontaria harvests the whole Semantic Web

Excluding forbidden content, following robots.txt, and ignore content which appears too big or is otherwise problematic.

IPT: 10d

Actors

The "actors" are simplifed roles which people or teams may play in their use of Ontaria. Any real person or group is likely to behave like one or more actor, as they solve a variety of real-world problems and as their tasks change over time. An awareness of the actors can help generate and clarify use cases.

Software Developers

Systems built to interoperate on the Semantic Web must be constructed in accordance with the appropriate selection of ontologies. A camera which records metadata in RDF will have hard-coded URIs based on the selected photography ontologies. Early-stage software developers may be trying to select ontologies or decide whether to produce their own; late-stage developers are heavily invested in the details of their chosen ontologies.

Direct Data Providers

Semantic Web direct data providers create RDF using tools which operate independent of specific ontologies, including text editors, scripting languages, or generic RDF editors. Like software developers, they need to know which ontologies to use to express knowledge, but their investment is in education and accumulated RDF content, not in software.

Indirect Data Providers

Indirect providers -- people using ontology-specific tools -- are insulated from the details of their ontologies. They care about which ontologies are used by their tools only to the extent that they care about standards compliance and potential interoperation. Their use of Ontaria is likely to be more limited than direct providers or software developers.

Ontology Providers

For creators, editors, maintainers, and publishers of ontologies meant for inter-organizational use, Ontaria offers a way to communicate with current and potential users. Providers vary in many respects (like the ontologies they provide), including who web-hosts what they publish, what support they offer, and their level of committment to maintain an ontology in accordance with stated goals and principles.

Direct Data Consumer

The direct data consumer uses RDF data without ontology-specific tools. These people may be debugging, reverse-engineering, or otherwise investigating something, using a relatively high level of RDF experise.

Indirect Data Consumer

The indirect data consumer relies on software which is hard-coded to work with particular ontologies. They might just use Ontaria to help them find software which can help them use some available data.