92 citations found. Retrieving documents...
Eric Brill. A Corpus-Based Approach to Language Learning. PhD thesis, University of Pennsylvania, 1993.

 Home/Search   Document Details and Download   Summary   Related Articles   Check  

This paper is cited in the following contexts:

First 50 documents  Next 50

Logic Form Transformation for WordNet Glosses and its Applications: .. - Rus (2001)   (Correct)

....the parser (parser should accept tagger s tag set) 31 it should have compatible tag sets One extra feature we watch for was to be able to easily modify the tagger to t our speci c data. There are many taggers available out there. We chose to use MXPOST tagger [34] and Brill s POS tagger [4][5][6] 7] as being the ones the most respect our guiding principles described before. MXPOST is a statistical tagger and it was trained on sections from Wall Street Journal corpus. MXPOST is a statistical tagger with a Maximum Entropy model that uses many contextual features (the word contains ....

Eric Brill. A Corpus-based Approach to Language Learning. PhD thesis, Department of Computer and Information Science, University of Pennsylvania, 1993.


Intelligent Key Prediction by N-grams and.. - Thanadkran.. (2001)   (Correct)

....key button K, K is the key button, n is the length of the considering token. 3.3.2 Error Correction for Thai Key Prediction In some cases of Thai character sequence, the tri gram model fails to predict the correct key. To correct these errors, the error correction rule method proposes by Brill [1, 2] is employed. 3.3.2.1 Error correction Rule Extraction Only the prediction errors after applying tri gram prediction to the training corpus are considered to prepare the error correction rule. The left and right three keys input Language Identification Key Input Thai Key Prediction Output ....

Brill, E. (1993) A Corpus-Based Approach to Language Learning. Ph.D. Dissertation, University of Pennsylvania.


Towards an Intelligent Multilingual Keyboard System - Potipiti.. (2001)   (Correct)

....Works Language Identification Key Input Thai Key Prediction Output Eng Yes Thai 3.3.2 Error Correction for Thai Key Prediction In some cases of Thai string sequences, the trigram model fails to predict the correct key. To correct these errors, the error correction rules as in [1] and [2] are employed. 3.3.2.1 Error correction Rule Extraction After applying trigram prediction to the training set, prediction errors happen. The left and right three keys input around each error character and the correct pattern corresponding with the error will be collected as an error correction ....

Brill, E. (1993) A Corpus-Based Approach to Language Learning. Ph.D. Dissertation, University of Pennsylvania.


A Statistical Information Extraction System for Turkish - Tür (2000)   (Correct)

....reach an F measure of 92 in the test set. In fact the HCRC LTG system, described in the next section, is the proof of this hypothesis. Alembic In MUC 6 evaluations, MITRE participated with the Alembic system [Aberdeen et al. 1995] using transformation based error driven learning algorithm of Brill [1993]. Their performance was in middle 80s. RoboTag Bennett et al. 1997] used binary decision trees using C4.5 [Quinlan, 1986] for name tagging task in the RoboTag system. The decision tree decides whether it is a name boundary or not. They use features indicating semantic properties (like first ....

Brill, Eric 1993. A Corpus-Based Approach to Language Learning. Ph.D. Dissertation, Department of Computer Science, University of Pennsyl- vania.


Efficient Algorithms for Parsing the DOP Model - Goodman (1996)   (10 citations)  (Correct)

....17.33 79.20 5.97 Zero Cross Brack DOP P S 1.33 20.00 21.33 6.93 5.65 Exact Match DOP 58.67 68.00 9.33 63.33 3. 22 Table 2: DOP versus Pereira and Schabes on Bod s Data A few sentences were not parsable; these were assigned right branching period high structure, a good heuristic (Brill, 1993). We also ran experiments using Bod s data, 75 sentence test sets, and no limit on sentence length. However, while Bod provided us with his data, he did not provide us with the split into test and training that he used; as before we used ten random splits. The results are disappointing, as shown ....

Eric Brill. 1993. A Corpus-Based Ap- proach to Language Learning. Ph.D. thesis, University of Pennsylvania.


Mixed-Initiative Development of Language Processing.. - Robyn Kozierok Patricia   (Correct)

....system, Alembic [1,7] to new tasks: the Message Understanding Conferences 0VIUC5 and MUC6) and the TIPSTER Multi lingual Entity Task (MET1) See [6] for an overview and history of MUC6 and the Named Entity Task . The Alembic text processing system applies Eric Brill s notion of ru e sequences [2,3] at almost every one of its processing stages, from part ofspeech tagging to phrase tagging, and even to some portions of semantic interpretation and inference. While its name indicates its lineage, we do not view the Alembic Workbench as wedded to the Alembic text processing system alone. We ....

Eric Brill. 1993. A Corpus-Based Approach to Language Learning. Ph.D. thesis, University of Pennsylvania, Philadelphia, Penn.


With - Maynard (1996)   (Correct)

....whether the preposition attaches to the preceding NP or to the V. Since most prepositions attach to the NP, it may always make the choice to attach the preposition to the NP, but in the case of with, this would be a bad choice since with is an exception, attaching more commonly to the verb [Bri93] At the next level, a more comprehensive parser will employ some more sophisticated means of disambiguation, such as the encoding of semantic information, e.g. case frames 16 for verbs, or the use of co occurrence and frequency based techniques, which can improve a parser s performance by ....

....this kind of knowledge, recent work has shown that more superficial forms of knowledge can be used successfully. Various researchers have attempted to resolve prepositional phrase attachment ambiguities by means of statistical approaches involving semantic word classes, e.g. WAB 91] BPV91] Bri93] HR93] These probabilistic models produce the interpretation with the greatest likelihood of occurrence. 26 3.1 Rule Based Approaches Brill and Resnik [BR94] describe a rule based approach to disambiguation of prepositional phrase attachment, which uses information automatically extracted ....

E. Brill. A Corpus based Approach to Language Learning. PhD thesis, Dept. of Computer and Information Science, University of Pennsylvania, 1993. 96


A Statistical Information Extraction System for Turkish - Tür (2000)   (Correct)

....reach an F measure of 92 in the test set. In fact the HCRC LTG system, described in the next section, is the proof of this hypothesis. Alembic In MUC 6 evaluations, MITRE participated with the Alembic system [Aberdeen et al. 1995] using transformation based error driven learning algorithm of Brill [1993]. Their performance was in middle 80s. RoboTag Bennett et al. 1997] used binary decision trees using C4.5 [quinlan, 1986] for name tagging task in the RoboTag system. The decision tree decides whether it is a name boundary or not. They use features indicating semantic properties (like first ....

Brill, Eric 1993. A Corpus-Based Approach to Language Learning. Ph.D. Dissertation, Department of Computer Science, University of Pennsyl- vania.


Massively Parallel Distributed Feature Extraction in.. - Kuntraruk, Pottenger (2001)   (Correct)

....Speech Tagging After identifying fields of interest, our feature extraction algorithms perform part of speech tagging. The part of speech tagger is a rule based system for tagging English parts of speech. This system is based on the SemanTag system developed in [7] which in turn is based on [3] [4] [5] The tagger uses three levels of rule sets to determine the part of speech of each word, and tags words with their English part of speech tag, as specified in the Brown tagset [12] DT determiner IN preposition or subordinating conjunction NN noun singular or mass PP personal ....

E. Brill. A corpus-based approach to Language learning. PhD thesis, University of Pennsylvania, Department of Computer and Information Science, 1993.


Viterbi Beam Search with Layered Bigrams - Goblirsch (1996)   (Correct)

....applications have a manageable number of subgrammars corresponding to different, easily identified semantic concepts. However, in general, automatic parsing of corpora into phrases and clustering of phrases will be needed to design layered bigrams optimally, perhaps using techniques described in [2], 3] or [5] Our definition of a layered bigram is somewhat weaker than that of [11] because in our case the probabilities that are internal to a bigram node are independent of its predecessor nodes; there is no conditioning on the children of these nodes. We have implemented a recognition ....

Eric Brill. A Corpus-Based Approach to Language Learning. PhD thesis, University of Pennsylvania, 1993.


A Statistical Information Extraction System for Turkish - Tür (2000)   (Correct)

....reach an F measure of 92 in the test set. In fact the HCRC LTG system, described in the next section, is the proof of this hypothesis. Alembic In MUC 6 evaluations, MITRE participated with the Alembic system [Aberdeen et al. 1995] using transformation based error driven learning algorithm of Brill [1993]. Their performance was in middle 80s. RoboTag Bennett et al. 1997] used binary decision trees using C4.5 [Quinlan, 1986] for name tagging task in the RoboTag system. The decision tree decides whether it is a name boundary or not. They use features indicating semantic properties CHAPTER 7. NAME ....

Brill, Eric 1993. A Corpus-Based Approach to Language Learning. Ph.D. Dissertation, Department of Computer Science, University of Pennsylvania.


A Rule-Based Named Entity Recognition System for Speech Input - Kim, Woodland (2000)   (1 citation)  (Correct)

....generates rules automatically. The procedures are mainly divided into two parts; preprocessing, and automatic rule generation. The preprocessing steps will be explained in Section 2.1. Then the automatic rule generation steps, the general idea of which originated from Brill s part of speech tagger [3], will be described in Section 2.2. 2.1. Preprocessing In this system, an untagged training data le is passed through the initial NE recogniser. The system separates all punctuation marks from their adjacent words, and then treats these punctuation marks as words. Training data with initial NE ....

E. Brill. A Corpus-Based Approach to Language Learning. PhD thesis, University of Pennsylvania, 1993.


Training a Selection Function for Extraction - Lin (1999)   (6 citations)  (Correct)

..... Translation is provided by Systran, http: www.systransoft.com sum of all the scores of content bearing terms in the sentence. These heuristics are implemented in separate modules using inputs from preprocessing modules such as tokenizer, part ofspeech tagger [6], morphological analyzer, term frequency and tf idf weights calculator, sentence length calculator, and sentence location identifier. 3. COMPARING the EFFECTIVENESS of HEURISTICS Initially, we implemented for SUMMARIST a straightforward linear combination function, in which we specified the ....

Brill, E. 1992. A Corpus-Based Approach to Language Learning. Ph.D. dissertation, University of Pennsylvania.


Morfologicke Znackovani Ceskych Textu - Hladka (2000)   (Correct)

....have to be an element of the list of tags returned by the MA for the given word. That is why the purely subtag independent strategy is modified by the so called Valid Tag Combination (VTC) strategy. Rule based approach The supervised transformation based error driven learning method described in [Brill, 1993] is classified as corpus based; however, we have to stress that it employs not only a small annotated corpus but a large unannotated corpus as well. A pool of allowable lexical and contextual transformations is predetermined by templates operating on word forms and word tokens, respectively. A ....

....the tagging procedure. The demand for more training data in case of a tagger with morphological preprocessing becomes more intensive. RB STRATEGY For Czech, we take the rule based tagger as is (designed for English) i.e. with the prespecified lexical contextual templates of the following form ([Brill, 1993]) The strategy of a rule based tagger determines the usage of annotated and unannotated corpora. The annotated corpus is being split into two parts of equal size. The first of these parts is used for learning the rules to predict the most probable tag for unknown words (lexical rules) and the ....

E. Brill. A Corpus-Based Approach to Language Learning. A dissertation in Department of Computer and Information Science, University of Pennsylvania, Philadelphia, USA, 1993.


Learning a Lexicalized Grammar for German - Kübler   (Correct)

....Therefore their membership value is increased. If the parser, however, does not succeed in parsing the sentence, the learning component is called: 13 Kbler Learning Lexicalised Grammar for German . As a first step, every word in the sentence is tagged. The formalism used for tagging will be Brill s (1993, 1995) transformation based error driven tagger. Unlike other approaches to learning using constituent based grammars, this system does not use the wordclass information to restrict the roles, a word can play in the parse. Rather it takes this information as a starting point in the search for ....

Brill, E. (1993). A Corpus-Based Approach to Language Learning (Ph.D. thesis). Philadelphia: University of Pennsylvania, Department of Computer and Information Science.


A Corpus for Interstellar Communication - Atwell, Elliott (2001)   (Correct)

.... (mapping words into syntactic semantic sets or classes) e.g. Atwell and Drakos 1987, Hughes 1993, Finch 1993, Hughes and Atwell 1994, Teahan 1998) Part of Speech wordtagging (mapping word sequences onto wordclass tag sequences) e.g. Leech et al. 1983, Atwell 1983, Eeg Olofsson 1991, Brill 1993, Atwell et al. 1984, 2000a) Sentence structure analysis or parsing (mapping word and or word class sequences onto parses) e.g. Sampson et al. 1989, Atwell 1987, 1988, 1993, Black et al. 1993, Bod 1993, Briscoe 1994, Jelinek et al. 1992, Joshi and Srinivas 1994, Magerman 1994, O Donoghue 1993, ....

Brill E 1993 A Corpus-based approach to language learning. PhD thesis, University of Pennsylvania.


Computer-Assisted Enlargement of Morphological Dictionaries - Daciuk   (Correct)

....the canonical form to the base form. i says that i characters should be deleted from the end of the canonical form. No ending is appended. Such rules can be discovered using a variety of techniques 2 . An obvious choice would be to use the transformation based error driven learning ( 3] [2]) Patterns of transformations should be established, and then individual transformations learned by choosing the best scoring transformation in each step. 9] used Brill s technique on a lexicon. However, his goal was not to learn the associations between the endings and the corresponding ....

Eric Brill. A Corpus-Based Approach to Language Learning. PhD thesis, Department of Computer and Information Science, University of Pennsylvania, USA, 1993.


Identifying Phrasal Connectives In Italian Using.. - De Santis, Tamburini (2001)   (Correct)

.... if two words are syntactically and semantically different, then they will appear in different contexts, as suggested in Harris [1951] There are a number of studies that, starting from this hypothesis, have constructed automatic or semi automatic procedures for clustering words [Brill et al. 1990, Brill 1993, Brown et al. 1990, Martin et al. 1998] They examine the distributional behaviour of some target words, comparing the respective collocates using some quantitative measures of distributional similarity. The work we present is based on a method first introduced by Brill and Marcus [1992] ....

Brill, E. (1993). A corpus-based approach to language learning. PhD thesis, Department of Computer and Information Science, University of Pennsylvania, Philadelphia, PA.


A Context-Sensitive Model for Probabilistic LR Parsing of Spoken.. - Ruland (2000)   (Correct)

....errors and a growing number of annotators. Hence we had to develop a technique to improve the exact match rate particularly with regard to the following semantics construction process that depends on correct syntactic analyses to produce a correct semantic representation of the utterance. (Brill, 1993) applied transformation based learning methods to natural language processing, especially to part of speech tagging. He showed that it can be effective to let a system make a first guess that may be improved or corrected by following transformation based steps. We observed many systematical errors ....

Brill, E. A Corpus-Based Approach To Language Learning. PhD Thesis, Department of Computer and Information Science, University of Pennsylvania, 1993.


Exploiting Semantic Extraction for Spatiotemporal.. - Vilain, Hyland, Holland (2000)   (Correct)

....semantic interpretation by taking a nontraditional view of syntax. Syntax is approached as a stratified process. We share with other recent systems the bottom levels of the syntactic strata, namely part of speech tagging, and name finding, both of which we accomplish through rule sequences (Brill 1993). We next group core phrases into headed syntactic units, or chunks, to use Abney s (1996) term. Chunks (or properly, their heads denotations) are then placed into predicate argument relationshbips through a repertoire of grammatical relations, yielding an interpretation graph (Fero et al. 1999) ....

.... actively exploited by the currently running GeoNODE (being a work in progress GeoNODE is only now starting to use the downstream capabilities of Alembic) The Phrase Rule Parser The lower strata of our parser operate as rule sequence processors, following the processing strategy made popular by Brill s (1993) widely distributed part of speech tagger (we use a version of this tagger in Alembic) Generally speaking, a rule sequence processor, approaches its designated processing task as a rewrite operation on strings news stories in our case. The string is given an initial partial labelling by some ....

Brill, E. (1993). A Corpus-based Approach to Language Learning. PhD thesis, U. Pennsylvania.


Using Induced Rules as Complex Features in Memory-Based.. - van den Bosch (2000)   (Correct)

.... further research should focus on the scaling properties of the approach (including the scaling of the external ruleinduction algorithm) should investigate more and larger language data sets, and should seek comparisons with other existing methods that claim to handle complex features efficiently (Brill, 1993; Ratnaparkhi, 1997; Roth, 1998; Brants, 2000) 77 Acknowledgements The author thanks the members of the Tilburg ILK group and the Antwerp CNTS group for fruitful discussions. This research has been made possible by a fellowship of the Royal Netherlands Academy of Arts and Sciences (KNAW) ....

E. Brill. 1993. A Corpus-Based Approach to Language Learning. Dissertation, Department of Computer and Information Science, University of Pennsylvania.


Estimating HMM Topologies - Brants (1995)   (Correct)

....the Viterbi algorithm [ Viterbi, 1967 ] 2.2 Part of Speech Tagging The task of part of speech (PoS) tagging is the unique annotation of a word with a syntactic category, called part of speech or tag. This paper is concerned with statistical PoS tagging. For rule based approaches, see e.g. Brill, 1993 ] and [ Voutilainen, 1994 ] Let T be defined as the set of all tags, and Sigma the set of all words. In a statistical tagging task, one is given a sequence of words W = w 1 : w k 2 Sigma , and is looking for a sequence of tags T = t 1 : t k 2 T that maximizes the conditional ....

Eric Brill. A Corpus-Based Approach to Language Learning. Ph.D. Dissertation, Department of Computer and Information Science, University of Pennsylvania, 1993.


Using the Link Parser of Sleator and Temperley to.. - Richard Sutcliffe.. (1996)   (3 citations)  (Correct)

....observed. Thus changes is not a noun. The same applies for verbs, meaning that display is not a verb. not is not counted as a verb and neither are nominalisations like Scrolling . The number of nouns and verbs in each utterance was established by tagging the corpus with the Brill Tagger (Brill, 1993) and correcting the output by hand. We took category C to mean compound nouns only. Thus we did not consider compound verbs (such as write down ) in any phase of the analysis. In Phase I no compound analysis was allowed before parsing. However, LPARSER can analyse compound nouns quite well. In ....

Brill, E. (1993). A Corpus-Based Approach to Language Learning. Ph.D. Dissertation, Department of Computer and Information Science, University of Pennsylvania.


Probabilistic and Rule-Based Tagger of an Inflective.. - Hajic, Hladká (1997)   (2 citations)  (Correct)

....the tag to O2A if adding the prefix nej results in a word 3.2.2 LEARNING CONTEXTUAL CUES The second stage of training is learning rules to improve tagging accuracy based on contextual cues. These rules operate on individual word tokens. 4 We use the same names of files and variables as Eric Brill in the rule based POS tagger s documentation. TAGGED CORPUS manually tagged training corpus, UNTAGGED CORPUS collection of all untagged texts, LEXRULEOUTFILE the list of transformations to determine the most likely tag for unknown words, TAGGED CORPUS 2 manually tagged training ....

Eric Brill. 1993. A Corpus Based Approach To Language Learning. PhD Dissertation, Department of Computer and Information Science, University of Pennsylvania.


Disambiguation by Association as a Practical Method.. - Sutcliffe, Slater (1995)   (1 citation)  (Correct)

....to disambiguate the corpus. The results produced by each algorithm were compared with those indicated by the human subjects. These are summarised in Table 2. 3. 3 Experiment 2 : Corpus syntactically tagged In the second experiment we tagged the corpus for syntactic category using the Brill Tagger (Brill, 1993). This assigns a syntactic category to each word in the text with high accuracy. We then ran the Lesk, CVI 1 and CVI 2 models again, this time utilising the syntactic information to restrict possible word sense choices. This was accomplished by only considering word senses produced by the ....

Brill, E. (1993). A Corpus-Based Approach to Language Learning. Doctoral Dissertation, Department of Computer and Information Science, University of Pennsylvania.


Linguistic Annotation of Two Prosodic Databases - Wolters   (Correct)

.... 3 and converted to SAMPA for American English (Wolters, 1997) G was transcribed by a rule set implemented in the language P TRA (Stock, 1992) Each vowel symbol is followed by an indication of stress type, primary (1) secondary (2) or none (0) The AE POS tags were provided by the Brill tagger (Brill, 1993), which uses the Penn Treebank tagset (Santorini, 1990) The POS tags for Gwere derived from the Bonner Wortdatenbank (Brustkern, 1992) 5 The Phrase Layer There are two types of units, syntactic phrases and lists (tag: li ) Lists are frequently used in applications of text to speech ....

E. Brill. 1993. A corpus-based approach to language learning. Ph.D. thesis, University of Pennsylvania, Department of Computer and Information Science.


Treatment of Unknown Words - Daciuk (1999)   (Correct)

....words that cannot be found in a lexicon. It seems impossible to record all words of a living language in a lexicon, as a lexicon is static in nature, and a language is a living thing new words are coined continually. Another reason for nding words not present in the lexicon is the Zipf s law[Bri93] The Zipf s law states that the rank of an element divided by the frequency of occurrence is constant. e.g. in the Brown corpus, two percent of di erent words 1 account for sixty nine percent of the text. About seventy ve percent of di erent words occur ve or fewer times in the corpus. Fifty ....

....Later, they have been supplemented by statistical techniques (e.g. WMS 93] However, although the probabilities of di erent endings leading to their corresponding categories were calculated, the endings themselves were chosen manually. A revolutionary approach was proposed by Eric Brill ( Bri93] Bri95] The endings, as well as pre xes, are found by the program. Unknown words are rst tagged by a naive initial state annotator that assigns the tags proper noun or common noun on the bases of their capitalization. Then ve types of transformations are applied: Change the tag of an ....

Eric Brill. A Corpus-Based Approach to Language Learning. PhD thesis, Department of Computer and Information Science, University of Pennsylvania, USA, 1993.


Using Induced Rules as Complex Features in Memory-Based.. - van den Bosch   (Correct)

.... further research should focus on the scaling properties of the approach (including the scaling of the external ruleinduction algorithm) should investigate more and larger language data sets, and should seek comparisons with other existing methods that claim to handle complex features eciently (Brill, 1993; Ratnaparkhi, 1997; Roth, 1998; Brants, 2000) Acknowledgements The author thanks the members of the Tilburg ILK group and the Antwerp CNTS group for fruitful discussions. This research has been made possible by a fellowship of the Royal Netherlands Academy of Arts and Sciences (KNAW) ....

E. Brill. 1993. A Corpus-Based Approach to Language Learning. Dissertation, Department of Computer and Information Science, University of Pennsylvania.


Learning Unification-Based Natural Language Grammars - Osborne (1994)   (2 citations)  (Correct)

....ANLT example are either unary or binary, this is a plausible assumption to make. ffl The lexicon is complete. This is the strongest assumption and means that the system assumes that each word encountered is already tagged with its partof speech. Given the advent of robust lexical taggers (e.g. [24, 28, 8]) this assumption is commonly made by other workers. 8 ffl Competence and not performance grammars are learnt. Most other researchers learning grammar acquire performance and not competence grammars. This is because such researchers use inductive methods, with no model of grammaticality, and so ....

Eric Brill. A Corpus-Based Approach to Language Learning. PhD thesis, University of Pennsylvania, 1993.


A comparative evaluation of modern English corpus grammatical .. - Atwell, al. (2000)   (1 citation)  (Correct)

....to us and use similar underlying algorithms, they differ in significant ways. In some cases we do not have access to the tagging programs, and in one case (POW) the corpus was tagged manually by linguists. We decided to train a publiclyavailable machine learning system, the Brill tagger (Brill, 1993), to re tag according to all of the schemes we are working with. As the Brill tagger was the sole automatic annotator for the project we achieved greater consistency. The Brill system is first given a tagged corpus as a training set, to extract or learn a complex set of tagging rules for the ....

Brill, Eric. 1993. A Corpus-based approach to language learning. PhD thesis, Department of Computer and Information Science, University of Pennsylvania.


Visualisation of Long Distance Grammatical Collocation.. - Elliott, Atwell, Whyte   Self-citation (Eric)   (Correct)

.... [11] instead used a 1 st order Markov or bigram model of tag coocurrence, learnable from a pre tagged training corpus (sampled from Brown) augmented with some hand picked longer context constraints when post editors thought these could improve accuracy [1,2] The widely used Brill tagger [3] uses constraint rules rather than a statistical bigram model, but these constraints are machine learnt from a pre tagged training corpus, so the system can learn different tagging schemes from different tagged training corpora [15] The ENGCG approach [10] requires an expert linguist to devise a ....

Brill, Eric. 1993. A Corpus-based approach to language learning. PhD thesis, Department of Computer and Information Science, University of Pennsylvania.


A Report of Recent Progress in Transformation-Based Error-Driven.. - Brill (1994)   (10 citations)  Self-citation (Brill)   (Correct)

....as well. 2 2. TRANSFORMATION BASED ERROR DRIVEN LEARNING Transformation based error driven learning has been applied to a number of natural language problems, including part of speech tagging, prepositional phrase attachment disambiguation, and syntactic parsing [Brill 92, Brill 93, Brill 93a] A similar approach is being explored for machine translation [Su et al. 92] Figure 1 illustrates the learning process. First, unannotated text is passed through the initial state annotator. The initialstate annotator can range in complexity from assigning random structure to assigning the ....

....such that a rule is allowed to make reference to parts of speech, words, and word classes, allowing for rules such as Change the tag from X to Y if the following word belongs to word class Z. This approach has already been successfully applied to a system for prepositional phrase disambiguation [Brill 93a] 5. UNKNOWN WORDS In addition to not being lexicalized, another problem with the original transformation based tagger was its relatively low accuracy at tagging unknown words. 10 In the start state annotator for tagging, words are assigned their most likely tag, estimated from a training ....

[Article contains additional citation context not shown here]

E. Brill 1993. A corpus-based approach to language learning. Ph.D. Dissertation, Department of Computer and Information Science, University of Pennsylvania.


Some Advances in Transformation-Based Part of Speech Tagging - Brill (1994)   (105 citations)  Self-citation (Brill)   (Correct)

....for 12 Note that this transformation will result in the mistagging of actress. The 17th learned rule fixes this problem. This rule states: change a tag from plural common noun to singular common noun if the word has suffix ss. 13 This learner has also been applied to tagging Old English. See (Brill 1993b) training and the next 150,000 words were used for testing. Annotations of the test corpus were not used in any way to train the system. From the 950,000 word training corpus, 350,000 words were used to learn rules for tagging unknown words, and 600,000 words were used to learn contextual rules. 148 rules were learned for ....

Brill, E. 1993b. A Corpus-Based Approach to Language Learning. Ph.D. Dissertation, Department of Computer and Information Science, University of Pennsylvania.


Comparing linguistic interpretation schemes for.. - ATWELL, DEMETRIOU.. (2000)   Self-citation (Eric)   (Correct)

....God HN s G Expressions for P example H for example A have M to I have to X (similarly for set up, as well as, so that, next to, Edit Copy, Drag Drop, Options. etc. 4. The multi tagger: a family of PoS taggers We trained a publicly available machine learning system, the Brill tagger (Brill, 1993), to re tag according to all of the schemes we are working with. As the Brill tagger was the sole automatic annotator for the project we achieved greater consistency. The Brill system is first given a tagged corpus as a training set, from which it extracts a lexicon and two sets of nonstochastic ....

Brill, Eric. 1993. A Corpus-based approach to language learning. PhD thesis, Department of Computer and Information Science, University of Pennsylvania.


Using Statistical Models To Predict Phrase Boundaries For.. - Synthesis Eric Sanders   (Correct)

No context found.

Eric Brill. A Corpus-Based Approach to Language Learning. PhD thesis, University of Pennsylvania, 1993.


Transformation Based Learning for Specialization of.. - Taffet, McCracken.. (2002)   (Correct)

No context found.

Brill, E. (1993). A corpus-based approach to language learning (Ph.D. Thesis). Philadelphia, PA: Department of Computer and Information Science, University of Pennsylvania. Available at: http://www.cs.jhu.edu/~brill/dissertation.ps


The Role of the HDDI Collection Builder in.. - Bader, Callahan..   (1 citation)  (Correct)

No context found.

E. Brill, "A corpus-based approach to Language learning", PhD. Dissertation, Department of Computer and Information Science, University of Pennsylvania, 1993.


Using Statistical Models To Predict Phrase Boundaries For.. - Synthesis Eric Sanders (1995)   (Correct)

No context found.

Eric Brill. A Corpus-Based Approach to Language Learning. PhD thesis, University of Pennsylvania, 1993.


Towards an Intelligent Multilingual Keyboard System - Potipiti..   (Correct)

No context found.

Brill, E. (1993) A Corpus-Based Approach to Language Learning. Ph.D. Dissertation, University of Pennsylvania.


Implementation Of Automatic Capitalisation - Generation Systems For   (Correct)

No context found.

E. Brill, A Corpus-Based Approach to Language Learning, Ph.D. thesis, University of Pennsylvania, 1993.


Using Multiple Sources of Information For Constraint-based.. - Tür (1996)   (Correct)

No context found.

E. Brill. A Corpus-Based Approach to Language Learning. PhD thesis, University of Pennsylvania, 1993.


Decision Lists For Lexical Ambiguity Resolution: Application to.. - Yarowsky (1994)   (55 citations)  (Correct)

No context found.

Brill, Eric, "A Corpus-Based Approach to Language Learning," Ph.D. Thesis, University of Pennsylvania, 1993.


A Statistics-Based Chinese Parser - Zhou   (Correct)

No context found.

Eric Brill (1993). A Corpus-BasedApproach to language Learning. Ph.D. thesis, University of Pennsylvania.


Automated Text Summarization in SUMMARIST - Hovy, Lin (1999)   (33 citations)  (Correct)

No context found.

Brill, E. 1993. A Corpus-Based Approach to Language Learning. Ph.D. diss., University of Pennsylvania.


ALLiS: a Symbolic Learning System for Natural Language Learning - Déjean (2000)   (Correct)

No context found.

Eric Brill. 1993. A Corpus-Based Approach to Language Learning. Ph.D. thesis, Department of Computer and Information Science, University of Pennsylvania.


Exploring the Statistical Derivation of Transformational Rule .. - Ramshaw, Marcus (1994)   (6 citations)  (Correct)

No context found.

Brill, Eric. 1993b. A Corpus-Based Approach to Language Learning. Ph.D. thesis, University of Pennsylvania.


TnT - A Statistical Part-of-Speech Tagger - Brants (2000)   (48 citations)  (Correct)

No context found.

Eric Brill. 1993. A Corpus-Based Approach to Language Learning. Ph.D. Dissertation, Department of Computer and Information Science, University of Pennsylvania.


Theory Refinement and Natural Language Learning - Dejean (2000)   (2 citations)  (Correct)

No context found.

Eric Brill. 1993. A Corpus-Based Approach to Language Learning. Ph.D. thesis, Department of Computer and Information Science, University of Pennsylvania.


ALLiS: a Symbolic Learning System for Natural Language Learning - Déjean (2000)   (Correct)

No context found.

Eric Brill. 1993. A Corpus-Based Approach to Language Learning. Ph.D. thesis, Department of Computer and Information Science, University of Pennsylvania.


Independence and Commitment: Assumptions for Rapid Training and.. - Hepple (2000)   (2 citations)  (Correct)

No context found.

Eric Brill. 1993. A corpus-based approach to language learning. Ph.D. thesis, University of Pennsylvania, Philadelphia, PA.

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