TweetFeel is a new web service by marketing research startup Conversition Strategies that combines real-time search for Twitter with sentiment detection algorithms.
The idea is for people to use TweetFeel to run search queries for products, celebrities, companies, brands etc. and thus get a notion of what the average Twitter user thinks of them in a matter of seconds.
TweetFeel evaluates real-time tweets about whatever search term the user has entered for positive and negative feelings, presumably taking into account words like ‘good’, ’sucks’, ‘great’, ’screw’, ‘love’ and whatnot. Search results flow down the screen as they are calculated as positive or negative, and the service offers an overall percentage number to indicate whether the majority of results are one or the other.
Individually, the results are evidently hit or miss, but in general there could be some interest from marketers to regularly cross-check references on Twitter for the company they work for, the products or services they market and the brands they represent.
Or you can just enter your name and see what people really think about you. In real-time.
would be great if it actually worked. tried two search terms and error on both of them
It doesn’t work for me too.. Also, I wonder why would any consumer want to use this service? Maybe, I would just give it a try to see how good it works. I would love to see a website where I can find what’s really popular among people, instead of finding sentiments.
I guess you might like http://bit.ly/Ty4es if you are looking for hot products among people.
Wow guys, this is really useful stuff. Definitely a positive technological development that will serve humanity for centuries.
I looked into this about a month ago (for product reviews). Fact of the matter is, we throw around emoticons as if they are meaningless.
To make the results worth anything you would need a trainable application, thousands of tweets and a few hundred volunteers. These volunteers would grade each tweet on a positive/neutral/negative scale and the backend, dynamic parser, would then learn more about the actual usage of language and emoticons on Twitter.
You can actually see the abuse of “emotional identifiers” directly within Twitter’s search (TechCrunch
). There are other minor inconsistencies as well, that would obviously be corrected through your service’s underlying technology, for instance 
is viewed as a negative emotion within Twitter search.
Ultimately, it was an investment in time and money well beyond my means and I ditched the idea.
The emoticons that were image-ified were the standard colon, close-paren smiley and the colon, capital P “tongue sticking out.”
Its really helpful to see peoples reactions to a brand you might be interested in, a movie you might want to see, or a link you might want to click on before you decide to do those things. TweetFeel seems like a welcome newcomer to that problem.
An even more organic approach would just be for Twitter users themselves to start adding tags into their tweets with their impressions, allowing any service to aggregate that information and then present it to other users. I suggest using * for impressions like # are used for topics. So if you just saw Ice Age 3, you could tweet “Just saw #IceAge3, thought it was *good.” Then a service could come along and aggregate those tweets into:
IceAge3
Top Impressions:
cute (211), good (112), funny (97)
You could do the same thing for links so that you could see a list of the most tweeted links recently and what everybody thought of them.
See http://microsyn...Impression-Tags for more.
sucks balls
less than impressive
We have been hit with some pretty heavy loads of traffic since this morning and this is causing issues with Twitter rate limits. We are working on fixing this and should have a solution soon.
Sorry. Entered the wrong twitter username. Here’s ours: @tweetfeeldotcom
This is not an innovative product. My former employer Techrigy, for example, has been doing sentiment analysis on Twitter (and all other social media) for over two years with their SM2 monitoring software. They now offer analysis of emotional responses (anger, frustration, greed, etc.) in addition to positive and negative sentiment.
It is important to bear in mind that there is no accurate automated sentiment analysis because software algorithms don’t understand irony, context, etc. The only service remotely close to accuracy uses humans to read and mark the results, a very costly proposition.
As Martin notes, assessing sentiment on Twitter is not new. Sysomos has two products – MAP and Heartbeat – that provides sentiment (as well as many other social media analytics features) about Twitter.
Mark
Isn’t this also what summize used to do before being acquired by twitter? There is a page here but it doesn’t seem to work anymore:
http://search.t...r.com/sentiment
summize was doing great before twitter bought them. since then, no innovation at all. sad.
Kind of interesting. Would be better if you could scroll down through the feed. At least on my browser, as soon as a tweet was below the browser window, there was no way to scroll down to see it.
Trying it out! Questions is can all human sentiments be defined by various emoticons and words? Given the field they are trying to address it would need some serious programming effort to ensure relevant and meaningful results!
There’s another interesting twitter based app worth a look http://www.tweetrsvp.com for questions and offers…
I think this is a valuable service.
RT @verbomania: Try saying “I Love microsoft” without grinning like an idiot – its impossible!
That was marked as a green smiley face.
looks interesting .
Looks like an indirect Twitter search engine, if you ask us
summize was doing great before twitter bought them. since then, no innovation at all. sad.
P.S. – Sorry, forgot to tell you great post!