About us

We are a small consulting firm based in London, UK, specialized in the application of computational linguistics to real life problems. We are a group of academics, with extensive background and research experience in linguistics, computer science, and computational linguistics, and experience in programming and managing implementations.

One of the biggest obstacles in academic research is to find actual data to work with. We are therefore very excited to bring our theoretical experience and programming knowledge to new kinds of data sets, and we pride ourselves in providing practical, efficient solutions that are useful for our clients.

We believe the most important foundation for a new project is to understand the purpose. We like to start a project with thorough data analysis and requirements analysis — after all, it’s no use developing a tool if it doesn’t work with your data, and even less use if in the end it doesn’t do what you thought it would.
We design efficient, state-of-the-art software systems to meet the established requirements — using off-the-shelf and open source components where possible, or by developing application-specific infrastructure where necessary. We can also manage and train your programmers to do the development and long-term maintenance of the system, so that you can have complete control of the implementation.

We are interested in hearing about new opportunities, so do get in touch if you have data that needs to be analysed, understood, or converted to language.
We specialize in a wide range of linguistic applications, from named entity extraction to natural language generation, and if we don’t have any expertise in a specific area, we are always happy to refer you to someone who knows more than we do.


We have expertise in most areas of computational linguistics that works with textual data, including

- parsing (sentences, structured documents or discourse)
- natural language generation
- ontology extraction
- named entity extraction
- sentiment analysis
- any kind of text processing including sentence segmentation, morphological analysis and generation, part of speech tagging, normalization

So far we haven’t worked with speech or images.

We have programming experience in Haskell, Perl, Python, Unix/Linux shell scripts, and work on MacOS, Linux or Windows operating systems.

We help you solidify requirements, design natural language processing pipelines, find open source tools to meet your needs or help you develop your own tools.


Eva Banik

Eva has a dual background in linguistics and computer science. Her first undergraduate degrees were in computer science and English from the University of Szeged in Hungary. She then went on to do a masters degree in linguistics at the University of Pennsylvania. Throughout her studies at UPenn, Eva has worked as a part time programmer/researcher and later as a programmer analyst at the Linguistic Data Consortium in Philadelphia on various projects (Open Archives Initiative, Less Commonly Taught Languages, XTAG). She moved to the UK in 2006 to do a PhD in computer science at the Open University. Her dissertation was on natural language generation: she designed and implemented an efficient and minimalistic generation architecture. Her implementation was built on top of the GenI surface realizer, which is how she got to know Eric. Before joining forces with Eric, Eva has worked on a named entity extraction project for Lexis Nexis.

Eric Kow

Eric is a long time Haskell programmer and overall computer geek. He has a PhD in computer science from the University of Henry Poincare in Nancy, France, where he worked on the TALARIS project. His PhD project was optimizing a surface realizer for natural language generation, and the results of his research are implemented in GenI. After his PhD, Eric moved to the UK, where he has worked as a research fellow at Brighton University on various projects (Prodigy,Generation Challenge 2010, 3D-Coform). In his spare time, Eric maintains the darcs revision control system and tries to encourage computational linguists to use more Haskell.


Natural language generation for HALO

We have worked with the Artificial Intelligence Center at SRI International on designing a full end-to-end natural language generation infrastructure to produce English texts from a biology ontology. Our work was part of project HALO, funded by Vulcan Inc and implemented as part of AURA.

KBGen 2013 – natural language generation from knowledge bases

We designed and ran a natural language generation challenge from knowledge bases in 2013. The challenge was built on data from the AURA knowledge base and the task was to generate fluent sentences from a set of triples. We provided a set of inputs in the form of triples and lexical items for the challenge, as well as the description of the meaning of relations in the knowledge base. For more information visit the KBGen website at kbgen.org where the data sets for the challenge are also available for download.


  • Eva Banik, Claire Gardent and Eric Kow (2013)  The KBGen challenge. In:Proceedings of the 14th European Workshop on Natural Language Generation, Sofia, Bulgaria.  pdf
  • Eva Banik, Eric Kow and Vinay Chaudhri (2013) User-Controlled, Robust Natural Language Generation from an Evolving Knowledge Base.  In: Proceedings of INLG2013, the 14th European Workshop on Natural Language Generation, Sofia, Bulgaria.  pdf
  • Eva Banik, C. Gardent, D. Scott, N. Dinesh and F. Liang (2012)  KBGen – Text Generation for Knowledge Bases as a New Shared Task. In: Proceedings of INLG 2012, The seventh International Natural Language Generation Conference. Starved Rock, Illinois, USA. paper poster
  • Eva Banik, Eric Kow, Nikhil Dinesh, Vinay Chaudhri and Umangi Oza (2012) Natural Language Generation for a Smart Biology Textbook. In: Proceedings of INLG 2012, The seventh International Natural Language Generation Conference. Starved Rock, Illinois, USA. paper slides
  • Eva Banik, C.Gardent, L. Perez-Beltrachini (2011) Natural Language Generation meets the Semantic Web : Knowledge Capture (K-CAP), Invited Tutorial, June 2011, Banff (Canada). slides
  • Eva Banik (2010) A minimalist approach to generating coherent texts. Phd thesis, Department of Computing, The Open University pdf
  • Eva Banik (2009) Parenthetical Constructions – an Argument against Modularity. In: Proceedings of the Grammar Engineering Across Frameworks workshop at ACL-IJCNLP-09, Singapore, 2009 pdf bib slides
  • Anja Belz and Eric Kow (2009) System Building Cost vs. Output Quality in Data-To-Text Generation. ENLG, 2009-02-11 pdf
  • Eva Banik (2009) Extending a Surface Realizer to Generate Coherent Discourse. In: Proceedings of the Short Papers of ACL-IJCNLP-09, Singapore, 2009 pdf bib slides
  • Eva Banik (2008) An Integrated Architecture for Generating Parenthetical Constructions. In: Proceedings of the Student Workshop at ACL 2008, Columbus, Ohio. pdf bib
  • Eva Banik and Alan Lee (2008) A Study of Parentheticals in Discourse Corpora — implications for NLG systems. In: Proceedings of LREC 2008, Marrakesh pdf bib poster
  • Anja Belz, Eric Kow, Jette Viethen and Albert Gatt (2008) The GREC Challenge 2008: Overview and Evaluation Results. INLG, 2008-06-12 pdf
  • Albert Gatt, Anja Belz and Eric Kow (2008) The TUNA Challenge 2008: Overview and Evaluation Results. INLG, 2008-06-12 pdf
  • Eva Banik (2007) Generating Parenthetical Constructions. Technical report 2007/14, Department of Computing, The Open University. pdf
  • Eric Kow (2007) Surface realisation: ambiguity and determinism. PhD thesis, University of Nancy, France pdf
  • Claire Gardent and Eric Kow (2007)  A Symbolic Approach to Near-Deterministic Surface Realisation using Tree Adjoining Grammar. ACL 2007. pdf bib
  • Claire Gardent and Eric Kow (2007) Spotting Overgeneration Suspects. ENLG, June 2007 pdf bib)
  • Claire Gardent and Eric Kow (2006)  Three reasons to adopt TAG-based surface realisation. TAG+8, July 2006 poster

Contact us

Email us


give us a call on

+44 (0)770 483 1818

or talk to us in person:

Podium Level
Shakespeare Tower
London EC2Y 8DR