K. Church, W. Gale, P. Hanks, and D. Hindle. Using Statistics in Lexical Analysis. In Uri Zernik, editor, Lexical Acquisition: Exploiting On-Line Resources to Build a Lexicon, pages 115-- 164, 1991.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....latter examples, in which the distinction between recipient and object relative to the dative verb PAY is made explicit, the former cases in which the relation is implicit can be resolved. In contrast to previous work which addressed the identification of surface relations, i.e. SVO triples [2], in our work we address the acquisition of semantic re lations, focussing at the assigment of thematic roles. This task (i.e. tagging for acquisition) requires high reliability and so it relies less on statistical properties and more on deterministic local marking. In this paper we discuss a ....

K. Church, W. Gale, P. Hanks, and D. Hindle. Us- ing statistics in lexical analysis. In U. Zernik, editor, Lecical Acquisition: Exploiting On-Line Resources. Lawrence Erlbaum Associates, Hillsdale, NJ, 1990.

....Our work provides a formal framework for the idea, extends the number of uses to which the models can be put, and introduces more rigorous evaluations of the tasks. Researchers have studied lexical, phrasal, co occurrence and dependency relations that can be pulled out from spans of text [11, 5]. Pseudo relevance feedback methods find words that are related to the query terms to improve the effectiveness of retrieval [24, 19, 14] Those uses of statistics and others like them are similar to our building of ELMs in that they find related words. We know of no work that viewed the ....

K. Church, W.Gale, P.Hanks, and D.Hindle. Using statistics in lexical analysis. In Lexical Acquisition: Exploiting On-Line Resources to Build a Lexicon, pages 115--164, 1991.

....into systems properly. 2. Principle 2.1. Mutual information and difference of t score between characters Mutual information and t score, two important concepts in information theory and statistics. have been exploited to measure the degree of association between two words in an English corpus[4]. We adopt these measures almost completely here, with one major modification: the variables in two relevant formulae are no longer words but Chinese characters. Definition 1 Given a Chinese character string xy , the mutual information between characters x and y(or equally, the mutual ....

Church K.W., Hanks P., Hindle D., "Using Statistics in Lexical Analysis", In Lexical Acquisition: Exploiting On-line Resources to BuiM a Lexicon, edited by U. Zernik, Hillsdale, N.J.:Erlbaum, 1991

....machine readable thesauri to assist in semantic tagging of texts. 2 Background Currently available dictionaries do not provide a sufficiently reliable source of lexical knowlege for NLP systems. This has led an increasing number of researchers to look at text corpora as a source of information [8, 22, 9, 6, 3]. For example, Brent [6] de scribes a program which retrieves subcategorization frames from untagged text. Brent s approach relies on detecting nominal, clausal and infinitive complements after identifi cation of proper nouns and pronouns using predictions based on GB s Case Filter [16] e.g. ....

....program recognizes five subcategorizatlon frames built out of three kinds of constituents: noun phrase, clause, infinitive. 83 Lexical acquisition of collocational information from preprocessed text is now becom ing more popular as tools for analyzing corpora are getting to be more reliable [9]. For example, Basili et al. 3] present a method for acquiring sublanguage specific selecttonal restrictions from corpora which uses text processing techniques such as morphological tagging and shallow syntactic analysis. Their approach relies on extracting word pairs and triples which represent ....

Church, K and Gale, W. and Hanks, P. and Hindle, D (1991) Using Statistics in Lexical Analysis. In Le:ical Acquisition, Zernik, Uri, Ed. Erlbaum, Hillsdale, NJ.

....on the left side and cat, which appears on the right. ProbM)ility: cv = P(catl,cat) Modified mutual information statistics (MIS) P(cat, cat= e(cq, where 111allS dollar care. MIS is similar to mutual infronmtion used by Church to calculate semantic dependencies be tween words [1]. MIS is different from nutual in formation because MIS takes account of the posi tion of the word (left right) Let us couslder au example Segmentation: two possibilities, 1) indirect) tax) and (2) i (new) f (tyt)c) lJ (indircct) remain as mentioned in section 1. Category ....

K. W. Church, W. Gale P. Hanks, and D. Hin- dle. Using statistics in lexical analysis. In Lezcal Acquisitin, chapter 6. Lawrence Erlbaum Associates, 1991.

....evaluate the e#ectiveness of query reformulation when searching. It has been shown that query reformulation can improve the e#ectiveness of a query [5] though the focus was on the cognitive burden it places on the searcher. This work was inspired by Tables 8 and 9 of a paper by Church et al. [7]. That work describes how statistical occurrence patterns of words in text can be used to find lexically interesting items. One of those is to look for verbs that occur near a particular word in an interesting way (measured by mutual information) allowing someone to find what does a boat do or ....

.... with particular query word, but many more than 50 times overall) Given the large number of patterns, making this decision in that way does not seem unreasonable, though it would ultimately be preferable to use a measure such as mutual information to select and keep the more interesting patterns [7]. 4.2 Patterns to questions The patterns are interesting, but they are not appropriate for displaying to a user. Instead, we convert them into questions that disambiguate the query. For example, here are some patterns and corresponding questions for the query word party : JJ party:NN varieties ....

K.W. Church, W.Gale, P.Hanks, and D.Hindle. Using statistics in lexical analysis. In Lexical Acquisition: Exploiting On-Line Resources to Build a Lexicon, pages 115--164, 1991.

....model. The degree of noun n s being the value of slot r for verb v represented by a conditional probability. Another way of learning case slot patterns for a slot for a verb is to calculate the association ratio measure, as proposed in (Church et al. 1989; Church and Hanks, 1989; Church et al. 1991). The association ratio is defined as , 2.2) where n assumes a value from the set of nouns, v from the set of verbs and r from the set of slot names. The degree of noun n being the value of slot r for verb v is represented as the ratio between a conditional probability and a marginal ....

Church, Kenneth, William Gale, Patrick Hanks, and Donald Hindle. 1991. Using statistics in lexical analysis. In Uri Zernik, editor, Lexical Acquisition: Exploiting On-line Resources to Build a Lexicon. Lawrence ErlBaum Associates, Hillsdale, pages 115--164.

....score(choice i ) and selects the choice that maximizes the score. The PMI IR algorithm is based on co occurrence. There are many different measures of the degree to which two words co occur (Manning and Schtze, 1999) PMI IR uses Pointwise Mutual Information (PMI) Church and Hanks, 1989; Church et al. 1991), as follows: score(choice i ) log 2 (p(problem choice i ) p(problem)p(choice i ) 1) Here, p(problem choice i ) is the probability that problem and choice i co occur. If problem and choice i are statistically independent, then the probability that they co occur is given by the product ....

Church, K.W., Gale, W., Hanks, P., and Hindle, D. (1991). Using statistics in lexical analysis. In Uri Zernik (ed.), Lexical Acquisition: Exploiting On-Line Resources to Build a Lexicon. New Jersey: Lawrence Erlbaum, pp. 115-164.

....member from the set of alternative words that is most similar in meaning to the problem word. The unsupervised learning algorithm performs this task by issuing queries to a search engine and analyzing the replies to the queries. The algorithm, called PMI IR, uses Pointwise Mutual Information (PMI) [1, 2] to analyze statistical data collected by Information Retrieval (IR) The quality of the algorithm s performance depends on the size of the document collection that is indexed by the search engine and the expressive power of the search engine s query language. The results presented here are based ....

....that maximizes the score. The PMI IR algorithm, like LSA, is based on co occurrence [9] The core idea is that a word is characterized by the company it keeps [10] There are many different measures of the degree to which two words co occur [9] PMI IR uses Pointwise Mutual Information (PMI) [1, 2], as follows: score(choice i ) log 2 (p(problem choice i ) p(problem)p(choice i ) 1) Here, p(problem choice ) is the probability that problem and choice co occur. If problem and choice are statistically independent, then the probability that they cooccur is given by the product ....

[Article contains additional citation context not shown here]

Church, K.W., Gale, W., Hanks, P., Hindle, D.: Using Statistics in Lexical Analysis. In: Uri Zernik (ed.), Lexical Acquisition: Exploiting On-Line Resources to Build a Lexicon. New Jersey: Lawrence Erlbaum (1991) 115-164.

....letters are in the same case. This could be explained by the fact that, whereas previously the words hThati and hthati were considered different, now they count as one word so the overall vocabulary is smaller and an entropy measure should be correspondingly less. Interestingly, Church et al. [30] note that, with certain polysemous words, case can be useful in disambiguating the proper sense; for example, the word hbanki has at least two distinct senses river bank and money bank. The capitalised hBanki is much more often associated with the money bank sense of the word; so retaining ....

....fruitful in terms of improving language model performance and also instructive in terms of developing an understanding of linguistic behaviour, and how this might relate to the more formal grammatical descriptions of theoretical linguists. Not all computational linguists agree Church et al. [30] prefer to emphasis semi automatic investigations. Using word classes does not guarantee an improvement in language model perplexity, even if it is assumed that all of the parameters of the model were set optimally. For example, if a classification system is used which includes a separate class ....

[Article contains additional citation context not shown here]

K. Church, W. Gale, P. Hanks, and D. Hindle. Using statistics in lexical analysis. In Uri Zernik, editor, Lexical Acquisition : Exploiting On-Line Resources to Build a Lexicon, chapter 6, pages 115 -- 164. Lawrence Erlbaum Associates, 1991.

....rely on confidence sets at least conceptually. These approaches define the degree of association of a pair of elements (e.g. words) to be the value of a statistic from a standard test of independence, such as the t test or the Chi squared test of association in binary (2 2) contingency tables (Church et al. 1991). Such an approach defines association to be related to how surprising the frequency of occurence of the associated pair is compared to the frequency of its elements, or alternatively, how unlikely it is that the same distribution generates both the observations of the pair and the individual ....

Church, Ken, William Gale, Patrick Hanks, and Donald Hindle. 1991. Using statistics in lexical analysis. In Uri Zernik, editor, Lexical Acquisition: Exploiting On-Line Resources to Build a Lexicon. Lawrence Erlbaum Associates, New Jersey, pages 115--164.

....to provide the mathematical background for understanding the statistical approaches to natural language processing. First we briefly introduce the basic probability theory, then the mutual information and the T score measure for lexical probability are explained. The various estimation methods (Church and Gale, 1991) which preclude zero probability events are shown later. Finally, we describe the EM method (Dempster et al. 1977) which serves the basis for the probability estimation for HMMs and PCFGs. 4.1. Basic probability theory A very brief introduction to probability and statistical theory is presented ....

....It is defined as follows: t = E(P (x; y) Gamma E(P (x)P (y) q oe 2 (P (x; y) oe 2 (P (x)P (y) 66) where E(P(x,y) and oe 2 (P (x; y) are the mean and variance of the probability of seeing word x followed by word y. Interested reader can refer to (Church et al. 1989) and (Church et al. 1991). As pointed out by Ken Church in (Church et al. 1989) these probability measures can be used in many practical applications: 1. Enhancing the productivity of lexicographers in identifying normal and conventional usage. 19 2. Enhancing the productivity of computational linguists in ....

[Article contains additional citation context not shown here]

Church, K., Hanks, P., Hindle, D., & Gale, W. (1991). Using Statistics in Lexical Analysis. In Zernik, editor, Lexical Acquisition: Using on-line Resources to Build a Lexicon. Lawrence Erlbaum.

....of parses, and mapped them to uninAEected forms using a full form word list. The resulting list contained 84182 non zero frequency counts. The frequency matrix was reduced by elimi 1 The parser is described in Hindle (1983) and Hindle (1994) Similar verb object data is discussed in Church et al. 1991). 12 AIMS VOL. 4 NO. 3 1998 asset bond interest security share stake stock unit average bit cent foot mark pence point yen cost debt dividend price rate rating tax value acquire 16 35 5 77 87 29 19 1 1 2 buy 16 48 53 36 348 107 190 29 2 2 1 2 2 dump 1 3 2 10 10 hold 18 7 22 5 68 121 30 3 1 1 ....

Church, K. W., W. A. Gale, P. Hanks, and D. Hindle (1991). Using statistics in lexical analysis. In U. Zernik (Ed.), Lexical acquisition: exploiting on-line resources to build a lexicon.

No context found.

Church, Kenneth, William Gale, Patrick Hanks, and Donald Hindle. 1991. Using statistics in lexical analysis. In Zernik (ed.) Lexical acquisition: exploiting on-line resources to build a lexicon.

No context found.

K. Church, W. Gale, P. Hanks, and D. Hindle. Using Statistics in Lexical Analysis. In Uri Zernik, editor, Lexical Acquisition: Exploiting On-Line Resources to Build a Lexicon, pages 115-- 164, 1991.

No context found.

Church, K. W., W. Gale, P. Hanks, and D. Hindle. 1991. Using statistics in lexical analysis. Lexical acquisition: Exploiting on-line resources to build up a lexicon, ed. U. Zernik, 115-64. Hillsdale, NJ: Lawrence Erlbaum.

No context found.

Church, K. W., Gale, W., Hanks, P., & Hindle, D. (1991). Using statistics in lexical analysis. In Zernik Uri, (ed.). Lexical acquisition: Exploiting On-line Resources to Build up a Lexicon. Hillsdale, NJ: Lawrence Erlbaum, pp. 115--164.

No context found.

K.W. Church, P. Hanks and D. Hindle. 1991. Using statistics in lexical analysis. In U. Zernik (ed.), Lexical Acquisition: Exploiting one-line resources to build a lexicon, Erlbaum.

No context found.

Church, K, W. Gale, P. Hanks, and D. Hindle. 1991 "Using Statistics in Lexical Analysis," in Zernik (ed.) Lexical Acquisition: Exploiting OnLine Resources to Build a Lexicon, pp. 115-164, Lawrence Erlbaum Associates Publishers

No context found.

Church, K, W. Gale, P. Hanks, and D. Hindle. 1991 "Using Statistics in Lexical Analysis," in Zernik (ed.) Lexical Acquisition: Exploiting OnLine Resources to Build a Lexicon, pp. 115-164, Lawrence Erlbaum Associates Publishers

No context found.

Church, K.W., Gale, W., Hanks, P., and Hindle, D. Using statistics in lexical analysis. In Uri Zernik (ed.), Lexical Acquisition: Exploiting On-Line Resources to Build a Lexicon, pp. 115-164. New Jersey: Lawrence Erlbaum, 1991.

No context found.

Church, K. W., W. A. Gale, P. Hanks, and D. Hindle (1991). Using statistics in lexical analysis. In U. Zernik (Ed.), Lexical acquisition: exploiting on-line resources to build a lexicon.

No context found.

Church, K. W., W. A. Gale, P. Hanks, and D. Hindle (1991). Using statistics in lexical analysis. In U. Zernik (Ed.), Lexical acquisition: exploiting on-line resources to build a lexicon.

No context found.

Church, K. W., W. A. Gale, P. Hanks, and D. Hindle (1991). Using statistics in lexical analysis. In U. Zernik (Ed.), Lexical acquisition: exploiting on-line resources to build a lexicon.

No context found.

Church, K. W., W. A. Gale, P. Hanks, and D. Hindle (1991). Using statistics in lexical analysis. In U. Zernik (Ed.), Lexical acquisition: exploiting on-line resources to build a lexicon.

First 50 documents  Next 50

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC