Powerset – Toward semantic search in a closed ecosystem

Peter Jones Information Ecology

Powerset provides advanced natural language browsing of searched terms and topics in Wikipedia. It’s designed to handle conversational language entries, and the tool is a good start. Try it on a few simple searches (e.g., a name) which is simple, then throw something abstract at it. Like “sensemaking” or “design theory” and the gaps in Wikipedia show up quickly. Wikipedia does not search across all articles for close matching terms (their search is an article finder, not a browse view). So Powerset fills a real need for knowledge awareness as Wikipedia becomes a popular starting point for Q&A, student-level research, and scanning the current cultural repertoire for memes and conventional wisdom.

The Powerset model makes sense – semantic relevance is achievable in a closed ecosystem where you have some level of editorial control of the content. They also index Freebase, which is much less mature than Wikipedia, so Powerset’s indexing of the two services does not yet offer access into deep knowledge resources.

For reaching deeply into authoritative publications, and indexing qualified (institutional) servers using the FAST search engine, I like Elsevier’s Scirus. It now looks almost exactly like Google, which was the direction we steered it in 2001 when I advised on redesign. In my opinion it has lost some personality on the start page due to its recent facelift though. Scirus’ indexing and retrieval are very powerful – the browse experience is much more inviting than Google Scholar, and it accesses highly relevant content form multiple artefacts, not just citable articles, but research reports, lecture notes, online presentations, good stuff shared online by the same authors Google Scholar only cites. Scirus indexes a different a closed content ecosystem as well – based on validity (academic, institutional, verifiable publications) and not domain (Wikipedia or .edu sites) or authority.