Tài liệu Báo cáo khoa học: "Unsupervized Word Segmentation: the case for Mandarin Chinese" doc

... measure without the need for ﬁne-tuning the balance between the two. The evolution of the results w.r.t. word length isconsistent with the supervized cross-evaluation re-sults of the various segmentation ... characters). However, they opti-mize their parameter for each setting. We thereforeconsider that their system does take into account the level of processing which is performed on Latin char-acters ... the string by substracting the mean of the VBEs of the strings of the same length. Writing˜δh→(x) and˜δh←(x). The normalized VBEs for the string x, or nVBEs, are then deﬁned as follow (weonly...

Tài liệu Báo cáo khoa học: "Joint Word Segmentation and POS Tagging using a Single Perceptron" docx

... some partial wordsare “justiﬁed” as complete words by the current POSinformation. On the other hand, if partial words arenot given POS tag features, the correct segmentation for long words can ... 1: The perceptron learning algorithmuseful only for the POS “number word in the base-line tagger, is also an effective indicator of the seg-mentation of the two words (especially “”) in the joint ... model.3.2 The decoding algorithmOne of the main challenges for the joint segmenta-tion and POS tagging system is the decoding algo-rithm. The speed and accuracy of the decoder isimportant for the...

Tài liệu Báo cáo khoa học: Evolutionary relationships of the prolyl oligopeptidase family enzymes docx

... life forms and that the b-propeller domain has been part of the family for billionsof years. There are striking diﬀerences in the mutation ratesbetween the enzymes and POP was found to be the ... based on the initial 3D alignment. The neighbor-joining tree was constructed for the peptidase domains of the enzymes (corresponding to the pig POP residues 1–72and 428–710) and for the complete ... proposed that the b-propellerof P. furiosus POP does not perform the same function as the mammalian enzyme, i.e. the exclusion of large peptidesfrom the active site [25]. Clearly the role of the b-propellerhas...

Tài liệu Báo cáo khoa học: "Fast Decoding and Optimal Decoding for Machine Translation" doc

... of e ), (the fertility of the NULL word) , (the k French word produced by e in a), (the position of in f), (the position of the ﬁrst fertile word to the left of e in a), (the ceiling of the average ... f, then an optimaldecoder will search for an e that maximizes P(ef)1 The symbols in this formula are: (the length of e), (the length off), e (the i English word in e), e (the NULL word) , (the ... “spurious” French words.Next, we perform a word- for -word replace-ment of English words (including NULL) byFrench words, according to the table t(f e ).Finally, we permute the French words. In per-muting,...

Tài liệu Báo cáo khoa học: "Large linguistically-processed Web corpora for multiple languages" doc

... various documents from the annotated corpus, we decided to perform a furtherround of cleaning. There are two reasons for this:first, we can exploit the annotation to find otheranomalous documents, ... (dictionary definitionsof the word, top pages of companies with the word in their name), whereas combining more than twowords retrieved pages with lists of words, ratherthan collected text.Ueyama ... Web9http://www.ims.uni-stuttgart.de/projekte/corplex/TreeTaggerdata (ultimately, the documents displaying the sec-ond problem might be among the most interest-ing ones to have in the corpus!), for now we try toidentify the most problematic documents throughautomated...

Tài liệu Báo cáo khoa học: "Text Alignment in a Tool for Translating Revised Documents" docx

... materials in the draft. In such cases, in addition to the revised text, the tool copies into the draft both the relevant text from the old version and the relevant translation and marks them appropriately. ... sections in the TL text from the existing translation and update materials from the SL text, thereby reducing the effort required from the translator. This tool could substantially increase the ... and marks them appropriately. The translator then can decide whether there is a point in using any of the existing TL text in the final translation of the document. 6 Conclusions and Future...

Tài liệu Báo cáo khoa học: "Improving Word Representations via Global Context and Multiple Word Prototypes" pdf

... contexts for clustering word instances, which is used in the multi-prototype ver-sion of our model that accounts for words with mul-tiple senses.We evaluate our new model on the standardWordSim-353 ... via a joint training objective. The model learns word representations that better cap-ture the semantics of words, while still keeping syn-tactic information. These improved representationscan ... architecture which 1) learns word embeddings that better capture the se-mantics of words by incorporating both localand global document context, and 2) accounts for homonymy and polysemy by...

Tài liệu Báo cáo khoa học: "Enhanced word decomposition by calibrating the decision threshold of probabilistic models and using a model ensemble" pdf

... describe the process of word generation from the left to the right by alternately using two dice, the ﬁrst for de-ciding whether to place a morpheme boundary in the current word position and the ... Another way of classifying ap-proaches is based on the learning aspect during the construction of the morphological model. If the data for training the model has the same struc-ture as the ... We then performed a fur-ther analysis based on relative word positions andfound out that the calibrated PROMODES-H pre-dicted non-boundaries better for initial word posi-tions whereas the...

Tài liệu Báo cáo khoa học: "Learning Word-Class Lattices for Deﬁnition and Hypernym Extraction" doc

... ωjb0 otherwisewhere ωkaand ωjbare the a-th and b-th word classesof skand sj, respectively. In other words, the matching score equals 1 if the a-th and the b-thtokens of the two ... amonochromeJJNN2structurepicturedotNN3dataFigure 1: The Word- Class Lattice for the sentences in Table 1. The support of each word class is reportedbeside the corresponding node.mal number of misalignments between the two to-ken ... relativeinfluence of the two methods on the final performance is notdiscussed.• The DEFINIENS field (GF): it includes the genus phrase (usually including the hyper-nym, e.g., “a first-class function”);• The...

Tài liệu Báo cáo khoa học: "Learning Word Vectors for Sentiment Analysis" ppt

... probability to a document d using a joint distribu-tion over the document and θ. The model assumeseach word wi∈ d is conditionally independent of the other words given θ. The probability of a docu-ment ... on the average polarity of docu-ments in which the words occur.Given a set of labeled documents D where skis the sentiment label for document dk, we wish tomaximize the probability of document ... sentiment labelˆs. We can then improve our word vector φwto betterpredict the sentiment labels of contexts in which that word occurs. For simplicity we consider the case where the sen-timent label...

Xem thêm