binarymax 13 days ago

“Relevant search” by Doug Turnbull and John Berryman, published by Manning, is THE best book to get started with tuning search engines.

I’be been a search engineer for >10 years and this is always the first book I recommend.

  • softwaredoug 12 days ago

    Awe thanks Max <3

    • deanebarker 12 days ago

      Before I read your book, I thought, "I know all about search!"

      After I read it...

      "...I knew NOTHING about search."

      No book has ever knocked me off my pedestal so brutally and so thoroughly.

    • aliswe 12 days ago

      For a moment I thought you were Doug Cutting

ssn 13 days ago

Three reference textbooks are available openly:

* Introduction to Information Retrieval,

* Information Retrieval in Practice,

* Entity-Oriented Search,

Modern Information Retrieval is also a classic reference. Not openly available but some contents are (were?) available online. Their site seems to be down but the Internet Archive has a copy.

Additional resources here:


  • firebones 12 days ago

    I am biased, but building the Intro to Information Retrieval chapters in your favorite language, bit by bit, is really good to get the feel of the tradeoffs for index capabilities.

100k 13 days ago

At a general audience level, "Index" is on my list to read. It covers the invention of the index up to digital search engines.

"Introduction to Information Retrieval" is a textbook which is available online Here's a review:

Another textbook which IMHO is a bit lower level is "Information Retrieval: Implementing and Evaluating Search Engines". The book website is down for me right now, but you can find it on Amazon here:

Another commenter linked to "Relevant Search", which is great if you want to learn how to effectively use a search engine to improve relevance (as opposed to how to implement a search engine). It's old, but another book in that vein that was really helpful for me earlier in my career is Lucene in Action:

  • driscoll42 12 days ago

    Going to second the rec on "Index", it's a very understandable, well researched book that the general audience or even a skilled practitioner would enjoy.

brudgers 13 days ago

Not a book but Hellerstein’s CS186 from 2015 starting with Lecture 17 gave me a basic understanding (I think).


Also from that lecture series, the low level is always IO. One disk read tends to dwarf n^2 in-memory algorithms.

And IO is all about tuning caches and hardware for the specific structural relationships in the data, the way in which it is accessed, and the hardware everything runs on.

Good luck.

MonkoftheFunk 12 days ago

Hotz... Is that you... Trying to learn to improve Twitter search? ;)

fiedzia 13 days ago
  • arooaroo 13 days ago

    Manning also have a book on Lucene, the library that powers Solr and ElasticSearch. IIRC the book covered how Lucene actually works under-the-good and would therefore act as a good reference on the subject in general.

  • gardenfelder 13 days ago

    Taming Text is about building a question-answering system; it came out about the time Watson came online; it's not a plan, rather a cookbook of experiments using Apache products like Solr and OpenNLP, but is a great tutorial on how question answering works.

vdfs 13 days ago

Lucene in Action, good introduction to Lucene, which can be helpful to learn ElasticSearch (most used FTS these days)

  • _tom_ 13 days ago

    Lucene in Action covers Lucene 3.0, and is from 2010. Current version is 9.4.2. So much has changed.

cb321 12 days ago

It's all in the Nim programming language, but if you prefer reading code or running diffs then you might get a vague sense of (some) low level nuts & bolts from:

User23 12 days ago

Is there some better alternative to Knuth-Morris-Pratt or Boyer-Moore? Both can easily be adapted to regular expression matching and as far as I know there’s no faster algorithm that doesn’t do preprocessing.

unixhero 13 days ago

Just use Postgres fulltext Search, its good enough

  • johnthescott 12 days ago

    for postgres, i highly recommend the rum index over the core fts. rum is written by postgrespro, who also wrote core fts and json indexing in pg.
    rum handles +20mil pdf pages, interactively.
    • SPBS 12 days ago

      Pleasantly surprised that RUM is just a drop-in replacement for the built-in GIN index, you can still use Postgres' native FTS operations with it.

    • unixhero 12 days ago

      Sounds very interesting. Never heard of rum, thank you for suggesting it.