Human activity recognition has often prioritized low-level features extracted from imagery or vid... more Human activity recognition has often prioritized low-level features extracted from imagery or video over higher-level class attributes and ontologies because they have traditionally been more effective on small datasets. However, by including knowledge-driven associations between actions and attributes while recognizing the lower-level attributes with their temporal relationships, we can attempt a hybrid approach that is more easily extensible to much larger datasets. We demonstrate a combination of hard and soft features with a comparison factor that prioritizes one approach over the other with a relative weight. We then exhaustively search over the comparison factor to evaluate the performance of a hybrid human activity recognition approach in comparison to the base hard approach at 84% accuracy and the current state-of-the-art.
This multilingual resource contains corpora in which verbal MWEs have been manually annotated. VM... more This multilingual resource contains corpora in which verbal MWEs have been manually annotated. VMWEs include idioms (let the cat out of the bag), light-verb constructions (make a decision), verb-particle constructions (give up), inherently reflexive verbs (help oneself), and multi-verb constructions (make do). VMWEs were annotated according to the universal guidelines in 19 languages. The corpora are provided in the cupt format, inspired by the CONLL-U format. The corpora were used in the 1.1 edition of the PARSEME Shared Task (2018). For most languages, morphological and syntactic information – not necessarily using UD tagsets – including parts of speech, lemmas, morphological features and/or syntactic dependencies are also provided. Depending on the language, the information comes from treebanks (e.g., Universal Dependencies) or from automatic parsers trained on treebanks (e.g., UDPipe). This item contains training, development and test data, as well as the evaluation tools us...
How do we gauge understanding? Tests of understanding, such as Turing's imitation game, are n... more How do we gauge understanding? Tests of understanding, such as Turing's imitation game, are numerous; yet, attempts to achieve a state of understanding are not satisfactory assessments. Intelligent agents designed to pass one test of understanding often fall short of others. Rather than approaching understanding as a system state, in this paper, we argue that understanding is a process that changes over time and experience. The only window into the process is through the lens of natural language. Usefully, failures of understanding reveal breakdowns in the process. We propose a set of natural language-based probes that can be used to map the degree of understanding a human or intelligent system has achieved through combinations of successes and failures.
Introduction <br><br> BOLT English PropBank and Sense -- Discussion Forum, SMS/Chat, ... more Introduction <br><br> BOLT English PropBank and Sense -- Discussion Forum, SMS/Chat, and Conversational Telephone Speech was developed by the University of Colorado Boulder - CLEAR (Computational Language and Education Research) and consists of propbank annotation and verb sense disambiguation annotation on English discussion forum (DF), SMS/Chat and conversational telephone speech (CTS) data. <br><br> The DARPA BOLT (Broad Operational Language Translation) program developed machine translation and information retrieval for less formal genres, focusing particularly on user-generated content. LDC supported the BOLT program by collecting informal data sources -- discussion forums, text messaging and chat -- in Chinese, Egyptian Arabic and English. The collected data was translated and annotated for various tasks including word alignment, treebanking, propbanking and co-reference. Data <br><br> DF data was collected from the web using a combination o...
Human activity recognition has often prioritized low-level features extracted from imagery or vid... more Human activity recognition has often prioritized low-level features extracted from imagery or video over higher-level class attributes and ontologies because they have traditionally been more effective on small datasets. However, by including knowledge-driven associations between actions and attributes while recognizing the lower-level attributes with their temporal relationships, we can attempt a hybrid approach that is more easily extensible to much larger datasets. We demonstrate a combination of hard and soft features with a comparison factor that prioritizes one approach over the other with a relative weight. We then exhaustively search over the comparison factor to evaluate the performance of a hybrid human activity recognition approach in comparison to the base hard approach at 84% accuracy and the current state-of-the-art.
This multilingual resource contains corpora in which verbal MWEs have been manually annotated. VM... more This multilingual resource contains corpora in which verbal MWEs have been manually annotated. VMWEs include idioms (let the cat out of the bag), light-verb constructions (make a decision), verb-particle constructions (give up), inherently reflexive verbs (help oneself), and multi-verb constructions (make do). VMWEs were annotated according to the universal guidelines in 19 languages. The corpora are provided in the cupt format, inspired by the CONLL-U format. The corpora were used in the 1.1 edition of the PARSEME Shared Task (2018). For most languages, morphological and syntactic information – not necessarily using UD tagsets – including parts of speech, lemmas, morphological features and/or syntactic dependencies are also provided. Depending on the language, the information comes from treebanks (e.g., Universal Dependencies) or from automatic parsers trained on treebanks (e.g., UDPipe). This item contains training, development and test data, as well as the evaluation tools us...
How do we gauge understanding? Tests of understanding, such as Turing's imitation game, are n... more How do we gauge understanding? Tests of understanding, such as Turing's imitation game, are numerous; yet, attempts to achieve a state of understanding are not satisfactory assessments. Intelligent agents designed to pass one test of understanding often fall short of others. Rather than approaching understanding as a system state, in this paper, we argue that understanding is a process that changes over time and experience. The only window into the process is through the lens of natural language. Usefully, failures of understanding reveal breakdowns in the process. We propose a set of natural language-based probes that can be used to map the degree of understanding a human or intelligent system has achieved through combinations of successes and failures.
Introduction <br><br> BOLT English PropBank and Sense -- Discussion Forum, SMS/Chat, ... more Introduction <br><br> BOLT English PropBank and Sense -- Discussion Forum, SMS/Chat, and Conversational Telephone Speech was developed by the University of Colorado Boulder - CLEAR (Computational Language and Education Research) and consists of propbank annotation and verb sense disambiguation annotation on English discussion forum (DF), SMS/Chat and conversational telephone speech (CTS) data. <br><br> The DARPA BOLT (Broad Operational Language Translation) program developed machine translation and information retrieval for less formal genres, focusing particularly on user-generated content. LDC supported the BOLT program by collecting informal data sources -- discussion forums, text messaging and chat -- in Chinese, Egyptian Arabic and English. The collected data was translated and annotated for various tasks including word alignment, treebanking, propbanking and co-reference. Data <br><br> DF data was collected from the web using a combination o...
Uploads
Papers