Text Analytics Solutions
We bring to the table a strong understanding of, and experience in, machine learning algorithms, especially Bayesian models and can help with:
- Developing resources like gazetteers, lexicons, morphological dictionaries, and classifiers
- Assisting with feature engineering for classifiers and clustering.
- Creating models for open frameworks like Weka (.arff files).
- Creating rules for GATE or other rule-based systems.
- Creating corpora.
We offer a number of ready-made pluggable components that can be integrated into enterprise software.
These include:
The DAT is a product for the annotation of textual data by a distributed team.
It lowers the cost and improves the scalability of corpora annotation.
It can be used to create corpora for parsing (grammar annotations), transliteration, entity/relation/event extraction, and sentiment analysis.
Click here for a demo of the DAT.
Classifiers
Text classifiers (or text categorisers) can discover interesting properties in groups of words, be they sentences, paragraphs or pages of text. In other words, they can place a piece of text into one of a number of categories.
For example, most web email accounts now automatically mark unwanted email messages as 'Spam'. Classifiers are the Natural Language Processing tool used to perform that task.
We can develop classifiers for you, and we have a range of topic classifier products available for purchase.
Here are some of the kinds of classifiers we can develop for you or license to you:
- Topic classifiers can identify the theme of a news article, say Politics, Sports, Health or Entertainment.
- Language classifiers can identify the language of a piece of text.
- Sentiment classifiers can tell whether a comment is positive (appreciative) or negative (derogatory).
We have performed tests on some of these and can supply you with accuracy figures for your feasibility studies. We can also help you evaluate open-source frameworks for classification should they serve your needs better.
Spelling Correctors
Spelling correctors are tools that automatically correct spelling errors.
There are different spelling correction algorithms and appropriate approaches to spelling correction can be selected based on the type of error to correct, the language, and the amount of memory available on the device where the spelling corrector runs.
Spelling correctors can also be reconfigured as text input tools.
Please note that we do not currently provide help with spelling correctors for non-compete reasons.