Welcome to the documentation for Cafetière Text Analytics.
To use Cafetière, go to cafetiere system.
-
Recent Posts
Recent Comments
Archives
Categories
Meta
Sentiment Analysis in Cafetiere
Sentiment Analysis in Cafetiere is work-in-progress. As currently deployed, we have a dictionary-based annotator that identifies subjective terms from the MPQA lexicon.
We calculate sentiment scores (+ve or -ve) for sentences in the text, and a mean for the text as a whole.
What we plan to do in the future is as follows:
- Enhanced sentiment dictionary. The current lexicon comprises entirely one-word terms. There are many stock phrases that people use to express positive and negative attitudes.
- Context-sensitive sentiment analysis, using a Cafetiere rule-based analyser. This could enable us to handle such phenomena as the rhetorical structures that can be used to build up an overall judgement that is not just the average of its parts.
- Plug in 3rd party sentiment analysis components, where they can be made to fit the Cafetiere framework.
- Make annotations editable, so that Cafetiere users can submit their own judgements of subjective expressions, in order to develop trained annotators for subjectivity.
Posted in Sentiment Analysis
Leave a comment
About
The Cafetiere Text Analytics toolkit lets you carry out simple text analysis on your own collection of texts, without needing to install anything, since you can run it in your Web browser.
It is designed for researchers in the social sciences and humanities. It can support a growing collection of analysis resources, but initially, resources are being uploaded for named entity recognition, term recognition and sentiment analysis. As this site grows, you will be able to obtain additional analysis resources and contribute your own.
Posted in Text Analytics
Leave a comment