Natural Language Processing(NLP)

Many think that NLP is a relatively new concept in computer science.I strongly disagree with that. NLP application like search engine (Information Retrieval System) is the most important piece of application that we use everyday life.Yet this was just a beginning. Since the search engine we really have not seen any significant NLP application(though many improved version of translation system,recommendation system embedded into many applications).

But the proliferation of Social Media (Facebook,Twitter and bla bla) has generated new attention on NLP. People talking heavily on how this massive big data generated by social media applications will help us and our business.The answer lies in building intelligent system that has natural language processing capabilities.

Researches already taking place on opinion mining of twitter data i.e. extracting sentiments from a tweet.This has many uses.For example in the last US Presidential Election Barack Obama Campaign Team has done lot of social data mining,text mining job to gauge public sentiments.Many companies are using sentiment tracking of their product reviews generated from blogs,facebook and twitter.

NLP uses many of the statistical learning algorithm from training the systems. There are two categories of learning algorithms-supervised and unsupervised.Supervised learning algorithms are based on Bayes Theorem and Unsupervised algorithms are  based on Markov model.There are many interesting learning algorithm exist in the machine learning paradigm.

Soon or later we will be building intelligent web robots with NLP capability who can learn from a blog like this and increase the knowledge base.Then the question will come how this knowledge would be useful to human?



  

Web Analytic

Online business is taking its shape. People getting online more and more. And we think there is business here. Google thought so long time back. They have some amazing tools. One of them is Web Analytic (Urchin).The Web analytic is a very fascinating tool to view the access analysis of a website i.e. number of page views, some statistical analysis on the page views like conversion rate. I do not want to discuss about the KPI s of websites. Google analytic has most of the KPI statistics for a website. Its a free service and very convenient. Next obvious questions come why anyone would really ever want to pay for the website access analytic. There are many reasons.　One simple reason is you do not want Google to have your private data. Big market players will never want that. Even a midsized company will be reluctant to add Google tracking code in their website.Thats why companies like omniture making money. But any medium sized website can not afford the cost of having ominuture.Good tools for web analytics are costly.

But when I looked at the Analytic technology closely I was surprised by the simple logic behind the Google analytic tool. It just uses AJAX to collect the information about the page. AJAX may be complex as a whole but when it comes to Web Analytic technology its really simple. Recently I have made an experiment about this. For my experiment I have put some javascript code on the hyperlinks of the page. And I fired the onclick event to call my ajax function in the server. And instantly I got to know that someone has clicked one of the links. The logic is simple and easy. But the interface will be complex I guess.
I guess my next step of building a analytic tool will be a proper plan and proper database design. May be next time when I write in this blog I will have more authenticated information about the Analytic tool. For now I really want to go ahead with this project.

Machine Learning

Machine Learning is the area of computer science which is growing rapidly.Some people says that this is the next big thing to be happened.Well this kind of words always pour s down whenever a new kind of topic hit the web.Machine Learning is not something like web2.0 or other general web related topic.In order to make Machine Learning a success we have to think of semantic web.Semantic web has been coined long back by Tim Barnes Lee .Its a visionary step and requires a lot of effort.Developers and programmers need to be educated first about the semantic web.Once the semantic web vision is full filled we would probably see a lot of machine learning robots on the web which crawl the web for finding interesting data and continuously training themselves for decision making.This is really exciting and would give a new generation of software that integrates desktop and web.A simple example would be a RSS robot.This is just an idea of myself.My RSS robot would crawl many news site for interesting news and says me good morning with a pan of todays breaking news from some chosen news services when I open my note PC.And I teach my robot time to time what and what not to serve me.This idea is really fascinating for me and hope one day real machine learning robots will crawl around the web following tons of millions of data.That day is not probably too far away..