Sentiment Analysis java Library [closed]

Question

I have some unlabeled microblogging posts and I want to create a sentiment analysis module.

To do this I have try Stanford library and Alchemy Api web service but the result it is not very good. For now I don't want training my classifier.

So I would like to suggest me some libraries or some web services about that. I would prefer a tested Library. The language of this posts is English. Also the preprocessing has been done.

P.S.

The programing language that I use is Java EE

Menhir

Questions asking us to recommend or find a book, tool, software library, tutorial or other off-site resource are off-topic for Stack Overflow as they tend to attract opinionated answers and spam. Instead, describe the problem and what has been done so far to solve it. – Dichroic

P.S. microblogging data (twitter?) usually is so crappy that it's no wonder when the analysis tools don't work very well. – Dichroic

@Anony-Mousse maybe you are right. But the question maybe can be in category: "if your question generally covers… software tools commonly used by programmers then you’re in the right place to ask your question!" But If my question is really out of topic I want to apologize about that. – Menhir

P.S. My data contains some tweets and some comments from news Portals. In second case the result can be acceptable, working with a good tool. – Menhir

Your question isn't about how to use such a library. If it were an actual, specific programming question then it would be on-topic, and not prone to polling and personal opinion (!). See my answer below, for what you really can expect from sentiment analysis these days. 99% hype, 1% delivered. Unfortunately. Technically, Stanford is probably one of the best(!) – Dichroic

Answer 1

If you want a good sentiment analysis service and you don't want to train your own classifier, you have to pay for it. However, it's worth mentioning that don't exist perfect tools in this field. There aren't tools that guarantee 100% of accuracy in their analysis.

Having said that, a couple of months ago I played around with Semantria/Lexalytics. They have a straightforward Java SDK and a good accuracy on their sentiment analysis results.

Quadriga

Thank you for your Answer. I know about this tools and the their results. But I would like to use a tool that guarantee 70% -80 of accuracy in their analysis. Do you know any free tool because I don't know If I can use non free tool for now? – Menhir

I don't think there're free tools to accomplish this job. A couple of months ago I did a research about those tools and I don't remember finding tools that meet your requirements. I believe you have only two choices here. The first choice is paying for it. The second choice is training your own algorithm, using Google Predict or Mahout, for example. – Quadriga

Ok thank you. The second step is to training my algorotithm. For now I must use a tool. – Menhir

Answer 2

Sentiment analysis doensn't keep up with the hyped promises.

See e.g.

The Sad State of Sentiment Analysis
December 26, 2013 by Angela Hausman
http://www.hausmanmarketingletter.com/sad-state-sentiment-analysis/

Recent experiments suggest sentiment analysis data is LESS accurate than a coin toss (accuracy 50%). That’s really scary if your brand makes strategic decisions based on sentiment analysis.

...

While the tools accurately predicted between 60 and 80% of utterances, when neutral utterances were removed (80% of the utterances) the accuracy dropped alarmingly.

In other words, everybody is cheating on their benchmarks, and overfitting (e.g. tweets have tons of duplicates and near duplicates - retweets - if you include these, you are overestimating the real performance)

Dichroic

Answer 3

LingPipe is a free(as well as paid) tool available for Sentiment Analysis. http://alias-i.com/lingpipe/index.html

Main features include:

1. Sentiment Analysis

2. Named Entity Recognition

3. Clustering

4. Topic Classification

5. Language Identification

etc

Paul

Thank you for your answer. But I was searching for a tool only for sentiment analysis. Also It will be helpful to indicate if you have tried this tool for a microblog (tweets) dataset. – Menhir

I'm currently working on Twitter sentiment analysis itself for movie reviews. But it seems that Stanford NLP is NOT best suitable. So I'm trying to train the NLP. BTW the results seem to be negatively biased with the initial trained classifier! Also, Stanford NLP is based on "sentence". Where as Tweets contain more than one sentence. Also we'll need to correct spellings. Hard luck! – Paul

Answer 4

Here check SentiStrength: http://sentistrength.wlv.ac.uk/

They claim that it works with tweets.

Bribe