We will try to integrate the one-click system with the translation tools or a hybrid based on robust parsing if available. We can decide the final architecture after the deliverables. Thomas should help on this. Alternatively, we can set up an online demo that translates a (small) bunch of patents or raw text.
We have a complete version of the grammars that covers 57 different queries. Maria will give, by Wednesday, detailed feedback and instructions about any required update (either the language or the SPARQL)
Meritxell will create a makefile to easily compile the document. UGOT will write a description of the last version of the grammar. Ontotext will try to find meaningful examples and update the interface accordingly Aarne suggested to stress the different MT approaches as an executive summary: - hybrids systems are SMT -driven - query-language is grammar and ontology driven Also, include Erzsébet in the authors list
Cristina will complete the part related to hybridisation Krasimir is rewriting the robust parsing part Ramona will complete the part related to GF (Section 4) Shafqat will probably finish the experiments on robust parsing this week. The deliverable will be extended later with the analysis relatedto this experiments.
MT summit, demo track. All WP7 people involved. Since MT is the focus of the conference, we will present a general description of the prototype stressing the two approaches followed in the hybrid vs. the queries.
Journal paper on the hybrid: Cristina and all
Multi-word Units in Machine Translation and Translation Technology: Conference paper about lexicon? (Ramona et al )
MT Summit/Workshop on patents with the most relevant aspect seen in D5.3