Apache Tika - a content analysis toolkit | Hacker News Reader

Apache Tika - a content analysis toolkit

tika.apache.org

60 pointszerop7 comments

+1 for Tika. I've been using Tika in conjunction with SOLR for a job search engine I built (www.neekanee.com) and it's really useful for extracting text from PDFs and .doc files if you're not overly concerned with the structure of the document.

The Apache Tika™ toolkit detects and extracts metadata and structured text content from various documents using existing parser libraries. You can find the latest release on the download page. See the Getting Started guide for instructions on how to start using Tika.

I read that and I still don't get it, but I can't find any other information on the site. Can someone provide a specific example of what problem this project solves?

Think of it as trying to pull text out of PDF / Office Docs / etc into a form (text, usually) that can be fed to a search engine, catalog, or whatever else you'd like to do with it.

At the core, it is a library that extracts text from various document formats (PDF, xlsx, docx, pptx, etc.).

It is important to note, however, that Tika in itself doesn't implement the text extractors, but rather combines different different projects (e.g., Apache POI) into one simple interface "getText".

If you are performance-oriented, thought, you may find that you will find yourself using the individual parsers as you'd want different settings/options to maximize throughput.

We used Tika for extracting data from various complex Office documents and converting them to Linked Data in the Proggis project: http://bergie.iki.fi/blog/business_analytics_with_couchdb_an...

In addition to content extraction, Tika seems able to do some interesting things like figuring out the language of the document contents.

On a related note, we are trying out a site which should help use projects like Tika - I am curious what you guys think of the site: http://www.codemaps.org/c/Apache_Tika

I'd like to see a comparison with Apache UIMA.

Reply on news.ycombinator.com