Looking Inside the Library Knowledge Vault OCLC Research Works in Progress webinar recording available
How do we ascertain truth on the web? That’s a question being pursued by researchers at Google who have articulated a flow of data that generates discrete statements of fact from countless web sources, relates those statements to previously assembled stores of knowledge, and fuses these mathematically to identify which statements may be more “truthful” than others. They describe this assembly of scored statements as a “Knowledge Vault.”
As OCLC works with data from library, archive and museum sources, we grapple with the same question and similarly varying data. Though the number of statements made is smaller and there may be fewer conflicts, we benefit by taking a closer look at the Google Knowledge Vault idea, to see how it applies to a vault of library knowledge.
In this webinar Jeff Mixter and Bruce Washburn provided an update on how they are evaluating this idea, including:
- extracting simple statements about entities and their relationships from bibliographic and authority records,
- establishing a relevant score for similar statements provided by different sources,
- viewing the Library Knowledge Vault data using a prototype application,
- and testing how statements contributed by users of that prototype can find their way back to the Vault.
This was the fourth webinar in our Works in Progress: an OCLC Research Occasional Webinar Series. OCLC Research Library Partners are invited to participate in these webinars to talk about work happening in OCLC Research, as well as to tell us what they are working on that everyone should know about and what input would help them move forward. Recordings of these webinars are made publicly available afterward.