Microsoft: Rationalizing Is a Synonym for Good Enough Search

May 25, 2020

On May 16, 2020, Microsoft — the JEDI champions and the target of amusement for Google’s Action Blocks — updated its “Rationalizing Semantic and Keyword Search on Microsoft Academic” page. One notable change is references to everyone’s favorite pandemic and bandwagon for virtue signaling: Covid 19.

What’s Microsoft saying about its Microsoft Academic Search?

The write up points out that the four year old method for delivering “results that best matched semantically coherent interpretations of user queries, informed by the Microsoft Academic Graph (MAG)” is fixed up. I assume this means the fixing up which Longhorn required before it became semi ready for prime time.

Microsoft points out (mostly in a mist of misinformation) that the competitors just do keyword matches. I won’t repeat what I have written in my three Google monographs, the New Landscape of Search, and numerous columns and blog posts.

Well, Microsoft does allow some stupid, old fashioned, and hopelessly archaic keyword searching. The new search will avoid returning pages with null results or zero hits. FYI, gentle reader, learning there are “no hits” is high value information for many queries. Just ask someone running scientific, technical, engineering, and medical queries. Those quite specific searches with no hits are informationized payloads.

Keyword matching is now “rudimentary.” And what’s better? Okay, Boolean lovers who know how to formulate specific queries created after a reference interview by the light of an oil lamp in a damp cave in Eastern Europe:

To put it simply, we’ve changed our semantic search implementation from a strict form where all terms must be understood to a looser form where as many terms as possible are understood.

What’s this mean? Irrelevant, or at best tangential information. But without the explicit mechanisms of a faceted based search system. (Endeca, Endeca, why did you beat up on those who wanted to perform “guided navigation?” Are you wizards to blame?)

The write up presents some before and after queries. Guess what? You get more results, more to scan and review, and more time burned because the search system is being helpful.

Ah, no, thank you.

There is zero search system of which I know capable of “knowing” how to relax a query to provide the specific information for which I am looking. I prefer to formulate a query, scan, reformulate the query, scan, and hone my attention to the content object which in my judgment a useful nugget of information can be found.

Microsoft presents data and “distance” as evidence their new and improved system works. Better than sliced bread? For Microsoft search experts, the answer is a chorus of “yes indeeds.”

The result is another modern system which makes a person less skilled in retrieving “academic” information get a “good enough” answer.

Remember. This Microsoft outfit is going to be in the warfighting game. How does “good enough” information retrieval intentionally displaying content not directly related to the query meet the needs of an analyst in one of the more academic units of the Pentagon?

Oh, I bet this new system is not intended for that PhD. That individual uses a next generation information retrieval which provides specific tools to locate on point information.

Microsoft wants to be the search champion. Too bad it is emulating the king of irrelevant results and doing it without the payoff of massive advertising revenue.

Need academic information? Gentle reader, try iSeek, Qwant or Swisscows or your library’s online commercial databases. Include Microsoft’s offering, but supplement, analyze, and aggregate. You know like do research, not accept what the JEDI crowd offers up.

Stephen E Arnold, May 25, 2020


Got something to say?

  • Archives

  • Recent Posts

  • Meta