Context in word recognition

Hanson, Allen R.; Riseman, Edward M.; Fisher, Edward G.

doi:10.1016/0031-3203(76)90027-3

Cited by 47 publications

(13 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A variety of spelling-related applications have been proposed and examined: n-grams have been used to correct Morse code (McElwain & Evens, 1962); flag possible typing errors (Morris&Cherry, 1975;Zamora, 1980;Zamora et al, 1981); perform general spelling correction (Angell, 1983;Ullmann, 1977); aid in optical character recognition (OCR) classification (Hussain & Donaldson, 1974;Shinghal, Rosenberg, & Toussaint, 1978); correct spelling errors in OCR after classification (Cornew, 1968;Damerau, 1964;Thomas & Kassler, 1967); and to do both detection and correction of OCR spelling errors (Hanson et al, 1976;Hull & Srihari, 1982;Neuhoff, 1975;Riseman & Hanson, 1974;Vossler & Branston, 1964). An overview of spelling-related applications and results may be found in the excellent reviews by Suen ( 1979) and Kukich ( 1992).…”

Section: N-gramsmentioning

confidence: 99%

Highlights: Language- and domain-independent automatic indexing terms for abstracting

Cohen

1995

J. Am. Soc. Inf. Sci.

108

View full text Add to dashboard Cite

Section: N-gramsmentioning

confidence: 99%

Highlights: Language- and domain-independent automatic indexing terms for abstracting

Cohen

1995

J. Am. Soc. Inf. Sci.

108

View full text Add to dashboard Cite

“…Now note that the loss function of (6) can be expressed by (7) L(zm, tm) == (10) 1 otherwise the loss function defined by (9) is called the total error loss function. It has loss equal to the number of incorrect assignments.…”

Section: Statement Of Problemmentioning

confidence: 99%

2. The decision-making context

1996

Whither Civil Engineering?

View full text Add to dashboard Cite

Abstract-From a Bayesian decision theoretic framework, we show that the reason why the usual statistical approaches do not take context into account is because of the assumptions made on the joint prior probability function and because of the simplistic loss function chosen. We illustrate how the constraints sometimes employed by artificial intelligence researchers constitute a different kind of assumption on the joint prior probability function. We discuss a couple of loss functions which do take context into account and when combined with the joint prior probability constraint create a decision problem requiring a combinatorial state space search. We also give a theory for how probabilistic relaxation works from a Bayesian point of view.

show abstract

“…Sometimes, the post-processing algorithms and the actual OCR are integrated to allow information exchange between the two. (Hanson et al, 1976) reported 2% word error rate and 1% reject rate without using any dictionary. By using an augmented dictionary and ignoring punctuation, (Sinha and Prasada, 1988) achieved 97% of word recognition with a Viterbi type algorithm.…”

Section: Dictionary Based Spelling Correctionmentioning

confidence: 99%