We address the issue of data sparseness problem in language model (LM). Using class LM is one way to avoid this problem. In class LM, infrequent words are supported by more frequent words in the same class. This paper investigates a class LM based on LSA. A word-document matrix is usually used to represent a corpus in LSA framework. However, this matrix ignores word order in the sentence. We pr...

The Duluth-WSI systems in SemEval-2 built word co–occurrence matrices from the task test data to create a second order co–occurrence representation of those test instances. The senses of words were induced by clustering these instances, where the number of clusters was automatically predicted. The Duluth-Mix system was a variation of WSI that used the combination of training and test data to cr...

A large number of news articles are published on the Web every day, and demand of discovering news articles on new/important topics has been growing. In this paper, we present a method for detecting characteristic words co-occurring with a target word (characteristic co-occurrence words) to help users find important topics related to the target word. The method divides news articles published i...

In this paper, a bottom-up, activation-based paradigm for continuous speech recognition is described. Speech is described by co-occurrence statistics of acoustic events over an analysis window of variable length, leading to a vectorial representation of high but fixed dimension called “Histogram of Acoustic Co-occurrence” (HAC). During training, recurring acoustic patterns are discovered and as...

• frequencies of occurrence of linguistic elements, which can be studied from two different perspectives: o how frequent are morphemes or words or patterns/constructions in (parts of) a corpus? This information can be provided in various different forms of frequency lists; o how evenly are morphemes or words or patterns/constructions distributed across (parts of) a corpus? This information can ...

In this paper, we present a Question Answering system called KUQA (Korea University Question Answering system) developed by using semantic categories and co-occurrence density. Semantic categories are used for computing the semantic similarity between a question and an answer, and co-occurrence density is used for measuring the proximity of the answer to the words of the question. KUQA is devel...

We present the use of grade correspondence analysis (GCA) in text mining. A sample of words extracted from 20 newsgroups has been linearly arranged according to concordance between their co-occurrence distributions. Words’ co-occurrence matrix, obtained using HAL (Hyperspace Analogue to Language) system and normalized to deemphasize too frequent terms, has been reordered by the GCA algorithm, i...

A comparison W~LS made of vectors derived by using ordinary co-occurrence statistics from large text corpora and of vectors derived by measuring the interword distances in dictionary definitions. The precision of word sense disambiguation by using co-occurrence vectors frorn the 1987 Wall Street Journal (20M total words) was higher than that by using distance vectors from the Collins English l)...

1. introduction in addition to simple verbs, persian employs a large number of complex predicates consisting of a preverbal element and a light verb. the preverbal element can be a noun, an adjective, an adverb or a preposition phrase, which combines with a verb to form a single syntactic predicate. persian complex verbs have attracted some researchers (e.g. folli, harley & karimi 2005, karimi ...