Báo cáo khoa học: "Learning to Identify Fragmented Words in Spoken Discourse" pot

... 3,137 fragmented words in our material, constituting 0.9% of the lexical tokens. The average sentence length in the corpus in 7.6 words. Interestingly, the average length of sen- tences containing ... test instance and a training instance stored in memory. When calcu- lating the differences, the features are ranked ac- cording to their importance in the classification task. A...

Ngày tải lên: 31/03/2014, 20:20

... sen- tences. Venugopal et al. (2009), in the string -to- tree set- ting, attempt to soften the data-fragmentation ef- fect of the matching constraint: instead of trying to find the single derivation with the ... dynamic programming human automatic string -to- string 198,445 142,820 max nested 78,361 64,578 tree -to- string 60,939 (78%) 48,235 (75%) string -to- tree 59,274 (76%) 46,548 (7...

Ngày tải lên: 07/03/2014, 22:20

... features in the joint model is 3,808. All models are trained using a Gaussian prior. 4.2 Data and baselines We divided the Kyoto Corpus (version 3.0) into the following three sections:  Training: ... often does) indicate multiple grammatical relations as in Ainshutain-ni akogareru "idolize Ein- stein" where ni marks the Object relation, and in To- kyo-ni sumu "li...

Ngày tải lên: 23/03/2014, 18:20

... outperforms Linear- Combination with any setting of (p 1 , p 2 ). The Pareto Frontier of PMO-PRO dominates that of Linear-Combination. This implies PMO is effective in optimizing towards Pareto hypotheses. 2. ... is to loop through the list L, keeping track of any dominant points. Given a dominant point, it is easy to filter out many points that are dominated by it. After suc- cessive ro...

Ngày tải lên: 19/02/2014, 19:20

... features to each Chinese token (characters or words) to reflect the likelihood of the token being part of the translation. We describe how to train a CRF model for identifying translations in more ... while requiring minimal human intervention to prepare the training data. 3 Method To find translations for a given term on the Web, a promising approach is automatically lear...

Ngày tải lên: 19/02/2014, 19:20

... set- ting it is situated in, motivating automated learn- ing approaches to acquiring meaning. Tradi- tional accounts of learning typically rely on lin- guistic annotation (Zettlemoyer and Collins, ... tend to be interpreted as going to the north and south of landmarks, respectively. Interestingly, our sys- tem tends to move in the opposite cardinal direc- tion, i.e. the agent moves...

Ngày tải lên: 20/02/2014, 04:20

... automated by natural language dialogue sys- tems, including information seeking functions, as in timetable or banking applications, but also more complex areas such as tutoring, health coaching ... similar to ours, constisting of: generation of (limited) exploratory dialogue data, using a train- ing system; creating an MDP from these data and the rewards assigned by the training users; of...

Ngày tải lên: 20/02/2014, 12:20

... this paper due to space constraint. 443 Line Line Line Line Line Line Line Line Line Line Line Line Line Line Line Line Line Line Line Line Line 1234567890123456789012345678901234567890123456789012345678901234567890 ... tables in Figure 1 to 3. Instead of resorting to an ad-hoc method to rec- ognize tables, we present a new approach in this pa- per that l...

Ngày tải lên: 20/02/2014, 19:20

Ngày tải lên: 22/02/2014, 03:20

... Q(s, a) to take into account words of the doc- ument in addition to state and action information. Conditioning Q(s, a) on all the words in the docu- ment is unlikely to be effective since only ... Reinforcement learning for mapping instructions to actions. In Proceedings of ACL, pages 82–90. S.R.K Branavan, Luke Zettlemoyer, and Regina Barzilay. 2010. Reading between the lin...

Ngày tải lên: 07/03/2014, 22:20