Tagging for Conceptual Frames
Red Hen has a major program for tagging for conceptual frames. She already annotates all of its English-language television caption data with conceptual frames from UC Berkeley's FrameNet project. Red Hen has, to our knowledge, by far the largest dataset thus annotated. But there are many further projects and a large, extended future for this research program. Many conceptual frames are not included in FrameNet; accordingly, Red Hen seeks to make a major fork of FrameNet that would be much more comprehensive. The tagging pipeline could be improved in various ways. Would you like to work on tagging for conceptual frames?
If so, write to
and we will connect you with a mentor.
How to start
See
German FrameNet and German Constructicon
Update the FrameNet tagger to Open Sesame
Frames listed in MetaNet (external)
Report on current research by FrameNet Brasil 2019-10-11:
FrameNet Brasil has been acting on 4 different groups of work:
1. In this line of work, FN-Br has expanded Berkley FrameNet's network of frames to Brazilian Portuguese (PT-Br) with the verification of the behavior of these frames in another language and, thus, promoting adjustments when necessary. Also part of this line is the creation of new frames for both Portuguese and English. FrameNet Brasil is a pioneer in the creation of frames from the Tourism and Sport domains.
2. FN-Br also uses this linguistic model for the development of non-linguistic applications. The two outstanding projects of this line are related with Tourism and Sports:
2.1 The World Cup Dictionary, a frame-based trilingual (Portuguese – Spanish – English) electronic dictionary covering the domains of soccer, tourism and the World Cup. Focused on human users, the final product was used by tourists, tourism professionals, journalists and the staff involved in the organization of the FIFA 2014 World Cup in Brazil.
2.2 The m.knob (Multilingual Knowledge Base) is an app based on a Conversational User Interface, GregBot, which acts as a virtual travel guide providing personalized recommendations of tourist attractions, events and activities. To do that, a Natural Language Understanding system uses the FrameNet Brasil database to derive semantic analysis of the sentences the user types during the conversation with GregBot and matches them to the descriptions of attractions in the database.
It also features a semantically grounded sentence translator and a Dictiopedia, a vocabulary of useful terms for tourism in Brazilian Portuguese, English and Spanish. The translations presented by the app are automatically generated via two different methodologies. For verbs and names that indicate events, they are calculated by a system that compares uses of these words in semantically annotated real texts stored in the FrameNet Brasil database. For names that indicate people, objects and places, they are automatically extracted from BabelNet, an open online database, and validated by the linguists in the project.
3. The FN-Br team is are also developing a Constructicon for Brazilian Portuguese, for which not only are new constructions, but also the alignment with the resources developed for other languages, such as English, German and Swedish.
4. FN-Br's latest foray has been in the field of multimodality, for which research is underway that investigates the relationship of language with other semioses, such as still images and videos, in a frame semantic approach.