Conference Paper

Dictionary supported generation of English text from Pitman Shorthand scripted phonetic text

Dept. of Studies in Comput. Sci., Mysore Univ., India;
DOI: 10.1109/LEC.2002.1182289 Conference: Language Engineering Conference, 2002. Proceedings
Source: IEEE Xplore

ABSTRACT The Pitman Shorthand Language (PSL) is a recording medium practised in all organizations, where English is the medium of transaction. It has the practical advantage of high speed of recording, more than 180 words per minute, because of which it is appreciably received. This recording medium continues to exist in spite of considerable developments in speech processing systems, which are not yet universally established. In order to exploit the vast transcribing potential of PSL a new area of research into automation of PSL processing is conceived. This paper describes the substitution of equivalent English words for the phonetic compositions of transcribed words, in the process of automatic generation of English text from a PSL document. Transcription is achieved by making use of two new types of dictionaries specifically developed and implemented for this purpose, one of them being a phonetic dictionary wherein the words are sequenced in phonetic order and the other being an extended conventional dictionary wherein the words are appended with additional details such as use domain, forms of verbs, etc. The proposed approach is tested with limited words in both dictionaries and is found to perform satisfactorily. However, the scope exists for addition of new words into these dictionaries.

1 Bookmark
  • [Show abstract] [Hide abstract]
    ABSTRACT: Innovative ways to rapidly input text becomes essential in today's world of mobile computing. The paper discusses the computer transcription of handwritten Pitman shorthand as a means of rapid text entry to pen-based computers, particularly from the aspect of linguistic post processing. Feature-to-phoneme conversion is introduced as the first stage of a text interpreter and the application of various production rules based on different pattern structures is discussed. It demonstrates that phoneme ordering is compulsory in dictionary-based transcription and the use of an approximate pattern-matching algorithm resolves the problem of recognition confusion between similar patterns. Experimental results are promising and demonstrate an overall accuracy of 84%.
    Frontiers in Handwriting Recognition, 2004. IWFHR-9 2004. Ninth International Workshop on; 11/2004
  • [Show abstract] [Hide abstract]
    ABSTRACT: This paper presents a detailed view of a novel solution to the computer transcription of handwritten Pitman's shorthand as a means of rapid text entry (up to 100 words per minute) into today's handheld devices with the use of a Bayesian network representation. Detailed design considerations of Bayesian network based shorthand outline models, including hypothesis of missing vowel components occurring in speed writing and unclear thickness and length of electrical pen-strokes are presented, along with graphical examples. Although Pitman's shorthand is written phonetically, our outline models are also based on low-level geometric attributes rather than phonetic attributes with the intention of coping with the unique features of handwritten Pitman's shorthand. The experimental results indicate an average accuracy of 92.86%, which is a marked improvement over previous applications of the same framework.
    Eighth International Conference on Document Analysis and Recognition (ICDAR 2005), 29 August - 1 September 2005, Seoul, Korea; 01/2005
  • [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we introduce a new approach to the computer transcription of handwritten Pitman shorthand as a rapid means of text entry (up to 100 words per minute) into today’s handheld devices, almost at the rate of speech. It is different from previous applications of the same framework from two aspects: – firstly, a novel idea of using geometric attributes other than phonetic attributes in the abstraction of a phonetic Pitman’s shorthand lexicon is proposed. Secondly, a Bayesian network representation for the organisation of shorthand-outline models is introduced, in which natural variability of Pitman shorthand is defined via different nodes and links. Using a probabilistic Bayesian network, the system shows a noticeable robustness not only in transcribing a variety of genuine handwriting, but also in estimating missing vowel components that may have been omitted in speed writing. The accuracy of the new approach (92.86%) is a considerable improvement over previous applications.
    Pattern Recognition and Data Mining, Third International Conference on Advances in Pattern Recognition, ICAPR 2005, Bath, UK, August 22-25, 2005, Proceedings, Part I; 01/2005