Innova is our permanent R&D line in language technologies. The trust of our clients and our research capacity drives this open research line and enables us to create innovative linguistic tools and applications.
Our most recent innovations are:
Apertium-Moses Hybrid Translators
Free hybrid translators for unrelated languages
- An alternative to Apertium (rule-based translator) for unrelated languages.
- Based on two free machine translation platforms: Apertium and Moses.
- Linguistic information is used to improve the performance of statistical machine translation.
Customisable, reliable and high quality
- Acquisition and filtering of semi-automatic corpora.
- Customisation of vocabulary and frequent expressions.
- Consistency of terminology, numeric data and punctuation e.g. capital letters.
- Wide variety of formats supported.
- More than 18 translators operating in working environments with excellent results.
- Modular translators which are being continuously improved.
Low-cost hybrid translation servers
- Training based on public, private and the client's own corpora.
- Customisation based on client's terminology or terminology extracted by us.
- Ready within 5 days as an on-demand service.
Extractium: Information Extractor and Classifier
Identification of relevant information in any text
- Fed by dictionaries and rules, customised by language.
- Available in Spanish, Portuguese and English.
- Customisable for specific uses and languages.
Uses: tag cloud, catalogues, intelligence
- Tag clouds for blogs and webs that identify individuals, organisations and places.
- Automatic data collection for catalogues, databases, etc. from official documents and other sources of information.
- Filtering and structuring of information for intelligence operations.
- Names of places and people.
- Addresses: postal, URL's, e-mails.
- Numerical expressions: telephone numbers, number plates, IP's.
Opinum: Opinion Analyser
Classification of positive and negative opinions
- Opinion classifier based on statistical techniques.
- Trained with corpora of opinion in a specific area.
- Cutomisable by environment and language.
Uses: feedback, analysis, decision making
- Filtering of feedback from users of blogs and websites.
- Analysis of products and services on the market.
- Strategic decision making.
Case study: banking sector, 81% accuracy
- Tested on opinions of financial products.
- Trained with a corpus of 9320 opinions in Spanish.
- Developed to distinguish between various degrees of positivity and negativity.
- Rate of accuracy: 81%.
Would you like to find out more about these technologies or about how we can meet new requirements arising in your working environment? Please fill in the form on the right explaining your requirements and we will get back to you shortly.