Ask HN: Books about full text search? | Hacker News Reader

Ask HN: Books about full text search?

232 pointssopromo31 comments

I would love to learn more about FTS at a very low level and I'm looking for books to read more on that topic. Any good suggestions ?

View on HN

“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.

https://www.manning.com/books/relevant-search

Awe thanks Max <3

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.

For a moment I thought you were Doug Cutting

Three reference textbooks are available openly:

* Introduction to Information Retrieval, http://informationretrieval.org/

* Information Retrieval in Practice, http://www.search-engines-book.com/

* Entity-Oriented Search, https://eos-book.org/

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:

* https://nlp.stanford.edu/IR-book/information-retrieval.html http://web.archive.org/web/20220708135205/http://grupoweb.up...

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.

At a general audience level, "Index" is on my list to read. It covers the invention of the index up to digital search engines. https://www.nytimes.com/2022/02/09/books/review-index-histor...

"Introduction to Information Retrieval" is a textbook which is available online https://nlp.stanford.edu/IR-book/ Here's a review: http://glinden.blogspot.com/2009/02/book-review-introduction...

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: https://www.amazon.com/Information-Retrieval-Implementing-Ev...

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: https://www.amazon.com/Lucene-Action-Second-Covers-Apache/dp...

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.

Managing Gigabytes

https://books.google.co.uk/books/about/Managing_Gigabytes.ht...

Old but good!

Came here to recommend Managing Gigabytes as well. People these days are managing far more than gigabytes but the fundamental ideas remain useful.

Check out the first review on the Amazon page. Norvig read it around the time he started at Google.

Not a book, but this paper from 2019 covers a lot of ground and reviews the different topics extensively: https://tonellotto.github.io/publication/fntir/fntir_main.pd...

Take a look at my post “Lucene: The Good Parts”—

https://blog.parse.ly/lucene/

The book mentioned there is Lucene in Action.

And then this YouTube presentation by a Lucene/Elasticsearch committer will give you a nice overview of some related algorithms—

https://youtu.be/eQ-rXP-D80U

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

Playlist https://youtube.com/playlist?list=PLhMnuBfGeCDPtyC9kUf_hG_Qw...

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.

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

https://www.manning.com/books/relevant-search

Also "taming text"

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.

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.

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

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

Check the literature of open courses on Text Retrieval. E.g. https://stanford.edu/class/cs276/

series of tutorials and comparisons that aim to teach a foundations about vector search:

https://vectorsearch.dev/

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: https://github.com/c-blake/nimsearch

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.

Stanford's NLP course:

https://www.youtube.com/playlist?list=PLoROMvodv4rOSH4v6133s...

123

456

Just use Postgres fulltext Search, its good enough http://rachbelaid.com/postgres-full-text-search-is-good-enou...

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.

    https://github.com/postgrespro/rum

rum handles +20mil pdf pages, interactively.

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.

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

Reply on news.ycombinator.com