NIST workshop on ontology evaluation
Bibliographic record
Abstract
The National Institute for Standards and Technology sponsored a workshop in October, 2007, on the subject of ontology evaluation.An international group of invited experts met for two days to discuss problems in measuring ontology quality.The workshop highlighted several divisions among ontology developers regarding approaches to ontology evaluation.These divisions were generally reflective of the opinions of the participants.However, the workshop documented a paucity of empirical evidence in support of any particular position.Given the importance of ontologies to every knowledge-intensive human activity, there is an urgent need for research to develop an empirically derived knowledge base of best practices in ontology engineering and methods for assuring ontology quality over time.This is a report of the workshop discussion and brainstorming by the participants about what such a research program might look like.ontologies: lack of a systematic method for evaluating ontologies, inadequate techniques for verification and validation, lack of standard methods for comparing ontologies, and paucity of real-world applications demonstrating effectiveness of ontologies.To address the issues above, a workshop was held at the National Institute of Standards and Technology on October 26 th and 27 th , 2007 to generate a research plan for the development of systematic methods for evaluating ontologies.The co-chairs of the workshop were Ram D. Sriram (National Institute of Standards and Technology), Mark A. Musen (Stanford University), and Carol A. Bean (National Institutes of Health).The topics for the workshop included the following: Representation.The language in which an ontology is expressed (its meta-language) should be used according to its intended syntax and semantics, to ensure that the ontology is properly understood by the user community and by computer-based tools.This topic addresses how to check that an ontology is using its meta-language properly. Accuracy.A well-constructed ontology is not very useful if its content is not accurate.This topic concerns methods to ensure that an ontology reflects the latest domain knowledge. Reasoners.An ontology can support automatic computation of the knowledge that is otherwise not obvious in the ontology.This topic addresses how to determine that automatically deduced information is consistent and valid. Performance metrics.Reasoners and other computational services are not very useful if they consume too many resources, including compute time.This topic concerns the bounds that users should expect from various kinds of computational services. Tools and Testbeds.Ontology evaluation is a complex task that can be facilitated by testing environments, graphical tools, and automation of some aspects of evaluation.This topic addresses computer-aided ontology evaluation. Certification.Ontologies that pass rigorous evaluation should be recognized by the community, to encourage the development and adoption of those of higher quality.This topic concerns the methods for official recognition of ontologies meeting high standards.Of particular concern is the role of social engineering to develop practices and tools that support the routine assessment and review of ontologies by the people who use them.The workshop had several presentations and breakout sessions.This report summarizes these presentations and breakout sessions.In our report of the discussions following each presentation, we use the abbreviation AM to connote an audience member, unless otherwise specified.Additional resources related to the workshop, including slides from each of the presentations are available at http://sites.google.com/a/cme.nist.gov/workshop-on-ontology-evaluation/Home/.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.081 | 0.093 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.002 |
| Bibliometrics | 0.010 | 0.006 |
| Science and technology studies | 0.006 | 0.004 |
| Scholarly communication | 0.016 | 0.017 |
| Open science | 0.005 | 0.009 |
| Research integrity | 0.006 | 0.007 |
| Insufficient payload (model declined to judge) | 0.011 | 0.008 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".