Member Avatar for mmona

Can a search engine be regarded as an expert system?
hey can anybody cud answer dis?

  • 3 Contributors
  • 2 Replies
  • 90 Views
  • 2 Days Discussion Span
  • Latest Post Latest Post by Mike Gale

Dani AI

Generated

Short answer for : not in the strict, classical sense. An "expert system" is a knowledge driven application with an explicit knowledge base plus an inference engine that draws and explains conclusions (Expert system). A "search engine" is primarily a large scale crawler, indexer and ranker that matches queries to documents and signals (Search engine).

Modern search platforms, however, contain components that behave like narrow expert systems. Features such as knowledge panels, featured snippets and structured answer modules combine curated facts, schemas and reasoning to produce single, phrased answers. Those pieces can be ruleful or fact based, but they sit inside a much larger retrieval and machine learning architecture rather than being a standalone rule based reasoner. See how structured knowledge is used in search via knowledge graphs (Knowledge graph).

Quick checklist to decide whether a system is an expert system:

  • explicit knowledge representation (rules, ontologies, facts) rather than only indexed text
  • an inference mechanism that derives conclusions from that knowledge
  • an explanation facility that shows how an answer was reached
  • narrow, evaluable domain performance (answers can be checked against expert criteria)

Use this checklist to evaluate specific search features. As and implied, the label depends on definitions and on which part of the engine is being examined.

Recommended Answers

All 2 Replies

Member Avatar for canadafred

Can a search engine be regarded as an expert system?

Sure, why not. An expert system of sorting through webpages and picking out the best results for a search. (at least I have the balls to take a stab at the question)

hey can anybody cud answer dis?

Yikes. Is there any way you can be more clear as to what your questions are? Really tough to answer the way they are presented. Whatdy'a say mmona, try again? I kinda' like this stuff.

Member Avatar for Mike Gale

Depends on your definition of ES. I see a wide variety of definitions some of them quite ridiculous.

My own definition gives a scale where -1 is "absolutely not an expert system", 0 is "heck I don't know" and 1 is "sure it's an ES".

I also believe that a system must actually do some good not just be a nice example to chatter about but never use in real life.

I also need to know how the system responds to small changes in inputs to judge how good it is.

Then each search engine is different. In the past some of them had block headed designs so were maybe negative scorers. I've not noticed a lot of continuous logics, which in my mind sorts the "nice try but no cigar" from the "good and has great potential".

So a pretty indifferent bunch among the search engines and they are getting worse. Partly through their own fault and partly not. There's a flood of rubbish content on the web (easy to publish) and they're losing out against that. Secondly some of them actively favour advertising in subtle ways, which is a direct move against being expert.

My best guess (without any serious work) there are probably search engines that are between say 0.01 and 0.15 of being a decent expert system. (Trend has been going backwards for many years.)

Edited by Mike Gale because: Make use of formatting available after sign-up but not during sign-up.
Be a part of the DaniWeb community

We're a friendly, industry-focused community of developers, IT pros, digital marketers, and technology enthusiasts meeting, networking, learning, and sharing knowledge.