Báo cáo khoa học: "Modeling Filled Pauses in Medical Dictations" docx

... Modeling Filled Pauses in Medical Dictations Serge)' V Pakhomov University of Minnesota 190 Klaeber Court 320-16 th Ave. S.E Minneapolis, MN 55455 pakh0002@tc.umn.edu Abstract Filled pauses ... in the CONTROLLED-FP-CORPUS. 620 4. Models The language modeling process in this study was conducted in two stages. First, a bigram model containing bigram probabilities of FP's in ... play a role in variability. My observations provide multiple examples where the doctors address the transcriptionists directly by making editing comments and thanking them. 3. Training Corpora...

Báo cáo khoa học: "Modeling with Structures in Statistical Machine Translation" pot

... tion we are more interested in cross-linguistic structures, similar to the case of using interlin- gua to represent cross-linguistic information in knowledge-based MT. To obtain structures that ... changing the threshold, we obtain a different number of phrases. The two operators are iteratively applied to the training corpus in alternative steps. This results in hierarchical phrases in ... the following two operators are used: . . Clustering: Clustering words/phrases with similar meanings/grammatical func- tions into equivalent classes. The mutual information clustering algorithm(Brown...

Báo cáo khoa học: "Modeling Topic Dependencies in Hierarchical Text Categorization" pot

... consti-tutes our MCC baseline.Regarding the reranker, we divided the trainingset in two chunks of data: Train1 and Train2. Thebinary classiﬁers are trained on Train1 and tested onTrain2 (and vice versa) ... (Child-free)Figure 6: Learning curves of the reranking models usingSTK in terms of MicroAverage-F1, according to increas-ing training set (child-free setting).0.3650.3750.3850.3950.4050.4150.4250.4350.4452 ... testTimeFRR trainingTimeFRR testTimeFigure 8: Training and test time of the rerankers trainedon data of increasing size.KPhighly improves on the baseline on child-freesetting by about 7.1...

Báo cáo khoa học: The hepoxilin connection in the epidermis docx

... ich-thyosiform erythroderma (NCIE, in layman’s termstranslating as a nonblistering, inherited, scaly redskin). An independent study later extended these find-ings following the identification of 17 ... isomerase(hepoxilin synthase)The genetic findings concerning 12R-LOX and eLOX3mutations in ichthyosis are intriguing from the bio-chemical point of view, partly because the eLOX3 pro-tein has been ... composed ofsphingosine in amide linkage to the carboxyl group ofa very long x-hydroxy acid (mainly C30,C32,orC34chain length), which in turn is esterified to the carb-oxyl group of linoleate;...

Báo cáo khoa học: "GENERATING PRECONDITION EXPRESSIONS IN INSTRUCTIONAL TEXT" docx

... proven useful in analyzing various kinds of conditions and circumstances that fre- quently arise in instructions. The analysis involves addressing two related issues: 1. Determining the range ... here the ter- minating condition). Finally, they may or may not be combined into a single sentence with the ex- pression of their related action (the issue of clause combining). Text generation ... manipulation of the evolving text structure, including the insertion of text structure nodes, grammatical marking of the nodes, textual ordering, and clause combining. Currently, the network...

Báo cáo khoa học: "HANDLING SYNTACTICAL AMBIGUITY IN MACHINE TRANSLATION" docx

... illustration, here we confine to prob- lems to be met with (i), and, more concretely, to such English strings containing Vin f. These strings are mapped onto Bulgarian strings containing da-construction ... convenient to distinguish two cases: Case A, in which to each syntactically ambiguous string in En- glish corresponds a syntactically ambiguous string in Bulgarlan, and Case B, in which to some ... strings which can be literally tran- slated into Bulgarian comprise,roughly speaking, the majority and the most common of strings to appear In real English texts. Informally, these strings...

Báo cáo khoa học: "Measuring Syntactic Difference in British English" docx

... inserted in a pathcontaining a leaf that is a leftmost sibling and a rightbracket is inserted in a path containing a leaf that isa rightmost sibling. The bracket is inserted at thehighest ... linguistic knowl-edge of the area being surveyed. These features,while probably lacking in completeness of coverage,certainly allowed a rough comparison of distance in all linguistic domains. ... bracketing ofthe previous two trees in detail. In the ﬁrst tree, withtwo B siblings, the ﬁrst path is A-B-p. Since p is aleftmost child, a left bracket must be inserted, at theroot in this...

Báo cáo khoa học: "Detecting Verbal Participation in Diathesis Alternations" docx

... The minimum description length principle is then used to produce a model and cost for storing the head noun instances from a training corpus at the relevant argument slots. Alternating sub- ... FOOD hyponym of OBJECT. Finding the best set of classes is key to obtaining a good preference model. Abe and Li use MDL to do this. MDL is a principle from in- formation theory (Rissanen, ... length. They provide a way of obtaining the ATCMS using the identity p(clv ) = A(c, v) × p(c). Initially a tree cut model is obtained for the marginal probability p(c) for the target...

Tài liệu Báo cáo khoa học: "Modeling Latent Biographic Attributes in Conversational Genres" pptx

... speakers usedfor training and 100 speakers used for testing, re-sulting in a total of 4062 conversation sides fortraining and 808 conversation sides for testing.4 Modeling Gender via Ngram ... training data and the ngram-based modelwas retrained on the remaining subset.Figure 2: Empirical differences in sociolinguistic featuresfor Gender on the Switchboard corpus6 Incorporating ... sociolinguistic and psycholinguistic pointof view of language understanding, but also froman engineering perspective, given the goal of pre-dicting latent author/speaker attributes in variouspractical...

Báo cáo khoa học: "Modeling Commonality among Related Classes in Relation Extraction" doc

... using the same feature set. 1 Introduction With the dramatic increase in the amount of tex-tual information available in digital archives and the WWW, there has been growing interest in ... of training examples for each class, since, in this case, a classifier learning approach can always learn a nearly optimal discriminative function for each class against the remaining classes. ... research. In the following, we will first introduce the perceptron algorithm in linear classifier learning, followed by the hierarchical learning strategy using the perceptron algorithm. Finally,...

Xem thêm