IBM TICL is something we made in our free time as a by-product of our research project earlier this year. It has now been released for free download and use on IBM Alphaworks. Expanding to IBM Tool for Interactive text Classification and Labeling, TICL is an interactive interface that enables you to create, validate, train, and refine a text ( Read more... )

The Holy Grail of text classification research is not to train a 100% accurate classification system, rather, it is to understand our own thought process better to produce consistent manually labeled data to train the system on.

In other news, my thesis is back from the two external examiners with good positive reviews and recommendations of ( Read more... )

Here is Clay Shirky's long article: Ontology is overrated that bashes ontologies and classification. Now, I understand the article is a generic rant, but some questions raised are valid questions that intelligent laymen would ask. Partly to dispel this, and partly to counter-argue, this is my wannabe practitioner's answer. The original article is ( Read more... )

This is a draft of the introduction section of my thesis synopsis/proposal/call-it-what-you-want. This is to solicit comments and feedback, here or by email, for language, style, content, organization. Please do read, especially since most of you are web-aware and computer science savvy enough.

( Click here - is intended to be readable )

This is partly speculative and partly straight from authoritative sources.

Dmoz and Yahoo are the most used directories on the web. They are manually maintained by volunteer editors and paid staff respectively. Dmoz's working is fully known to the public; it is based on discussion forums and consensus amongst volunteers. What Yahoo guys do and ( Read more... )

Tags in Blogs are definitely useful as they are an intuitive way of organizing entries. They can be looked upon as labels associated with documents. LJ has relatively recently introduced tags. Drupal, wordpress etc. have had them around for longer.

I am on this long term search for the holy grail blog aggregator. So, while Tejaswi finishes his ( Read more... )

For the record - ODP aka Dmoz is not built automatically. It is a manual categorization of web-pages into a complex hierarchy of categories. It stands, as of now, at 4.6M+ sites, 68k+ editors, 780k+ categories. The other directory of web-sites at Yahoo!, I suspect, is also fully manually built. The difference is that ODP is built by volunteer ( Read more... )