Báo cáo khoa học: "Setting Up User Action Probabilities in User Simulations for Dialog System Development" pot

... the trained and the manual simulations in generating corpus level dialog behaviors as well as in generating training corpora for learning dialog strategies. When com- paring on a dialog system ... time the human user takes an action in the dialogs in the testing data, the user simulations are used to pre- dict an action based on related dialog information from...

Ngày tải lên: 17/03/2014, 01:20

... In those techniques, frequency information of contextual words co- occurring in the monolingual text is stored and their similarity is measured across languages. The following gives a rough formalization ... that co-occur in the same sentence, or in a window of a few words. 2 In this query translation process, we also evaluated sim- ply consulting a bilingual lexicon instead of empl...

Ngày tải lên: 22/02/2014, 02:20

... spoken dialog. It also reports on the anno- tation that we performed in order to collect train- ing and test data for our machine learning experi- ments. The annotation also offered interesting in- sights ... options described above, so that no in- formation about interruption points or sentence boundaries was available during training or test- ing. With this setting, the classifier a...

Ngày tải lên: 22/02/2014, 02:20

... is often of interest in pro- tein science, especially in structural genomics where soluble proteins are a requirement for obtaining infor- mation about the 3D structure [10]. Several inherent parameters ... thereby each protein has more time to fold correctly and form a soluble protein before forming an insoluble protein precipitate by colliding with other recently translated proteins....

Ngày tải lên: 06/03/2014, 01:20

... the Escherichia coli FN3K PFKb Outer Membrane Protein Hypothetical protein Hydrolase YniCYniB Thermus thermophilus FN3KLMW-PTP Histidine kinase IndA protein Hypothetical protein GTP binding protein Cytophaga hutchinsonii FN3KLMW-PTP Fe uptake regulator Alkyl hydroperoxide reductase Hypothetical proteins Fig. 2. ... 4363 epsilon-amino group of lysine or to cadaverine (decarboxylated lysine) we...

Ngày tải lên: 07/03/2014, 05:20

... actually outperforms the in- domain classifier. This is possi- ble since the unlabeled data may contain information that the in- domain classifier does not have access to. At the beginning of Section ... 1000 most infor- mative features in both domains. In both cases, be- tween 85 and 90% of the informative features from one domain were not among the most informative of the other domain 4...

Ngày tải lên: 08/03/2014, 02:21

... constraint), no in- stances were created for that anaphor. Instances for nonreferential pronouns were added to the training data as described in Section 5.2. During testing, we select for each potential ... learning data for training and testing was created by pairing each anaphor with each of its compatible potential antecedents within a certain temporal distance (9 seconds for...

Ngày tải lên: 08/03/2014, 02:21

... transparent. We note in passing that although collapsing multilinear data-structures onto a single tier increases the likeliness of combinatorial explosion in process- ing when using the two-level ... used as in- put for speech synthesis. A distinguishing feature of the grammar used in the generator is the integration of sentence- level and word-level processing within the same...

Ngày tải lên: 08/03/2014, 06:20

... combining us- ing MERGE and using VOTE. We have chosen 4 baseline systems: (1) SM07; (2) the our base- line monolingual system using JCN for verb-verb comparisons (RelCont-BL), so as to distinguish the ... that 6 Using SemCor in this setting to augment WN does hint of using supervised data in the WSD process, however, since our approach does not rely on training data and SemCor is not...

Ngày tải lên: 17/03/2014, 00:20

... Behaviors: A Re-ranking Based Approach Yabin Zheng Zhiyuan Liu Lixing Xie State Key Laboratory on Intelligent Technology and Systems Tsinghua National Laboratory for Information Science and ... by leveraging user behaviors hiding in user records of Chinese input method. Our proposed method is motivated by the observa- tion that the more frequently two words co- occur in user...

Ngày tải lên: 20/02/2014, 04:20