1998 Volume 12 Issue 4
Published: 15 August 1998
-
Select all|
-
Review1998, 12(4): 2-9.In order to decrease the of single handwritten Chinese character recognition ,this paper discuesses the gerenal model of Chinese character recognition. Based on the discussion ,the paper proposer the model of multi - character recognition. To fully support multi - character recognition ,the paper supplies theory certification and experiment result s. Our experiment result s show that multi - character recognition can increase text recognition rate and is hopeful developing direction for handwritten Chinese character recognition.
-
Review1998, 12(4): 10-15.This paper int roduces the segmentation method of omnirange st ring. The iteration al2 gorithm of high accurate st ring directional determination is supplied which solved the direction correction for the short length st ring recognition ,and the ratiocination segmentation algorithm is also int roduced when the st ring is intersected with the line. We use the morphology analysis method to make complete recognition and ensure the integrity and the detectable property of the image information on the original image.
-
Review1998, 12(4): 16-22.Phrase recognition in fixed phrase table , with many applications , is an important research area of speech recognition. After an overview of phrase recognition , this paper proposes a Chinese phrase recognition method based on continuous speech recognition algorithm and phrase -tree const raint . This method get s high computing and searching efficiency by applying phrase -tree const raint information in continuous speech recognition algorithm. We first int roduce the beamsearch continuous speech recognition algorithm. Then the rest rictive phrase - t ree is presented with it s compact storage st ructure that improves the storage and access efficiency in recognition search. Finally , the experiment result s and discussion are given followed by a brief summary.
-
Review1998, 12(4): 23-30.This paper present s the methods for recognition of Chinese homophone words adopted by ICP system ,which is a Pinyin - Chinese characters conversion system. The methods take advantages of the context text function and realize the related information processing in indefinite range within a sentence. The part - of - speech , semantic meaning ,position and the f requencies of words are all considered as the evidence to recognize the Chinese homophone words by ICP system. ICP get s a high correct ratio of recognition of homophone words.
-
Review1998, 12(4): 40-47.An appropriate parsing to the sentences is necessary for sentence similarity calculation in EBMT. This paper put s forward a parsing approach —skeleton analysis ,which masters the main st ructure of the sentence through the determination of it s predicate head. Then discusses a st rategy to recognize the predicate head of the Chinese sentence according to the predicate head of the corresponding English sentence. The experimental result is satisfactory.
-
Review1998, 12(4): 48-53.By the definitions and descriptions of the notions ,that the level of title and paragraph ,the type of title in chinese text ,that indexing the writings - st ructure of chinese text is discussed in this paper. The concept of the regular chinese text and a tagging method on it s writings- structure are defined. On this basis ,an implementation technique and algorithm about automatic indexing for regular chinese text writings - st ructure are presented.
-
Review1998, 12(4): 54-60,9.We think over the specials of data ret rieving ,and build word segmentation dictionaries based on the lexicons of data ret rieve ; Then we propose the Different step - - Single scan Word Segmentation (DSWS) ,and analyze the time complexity of the segmentation algorithm. It is useful of this word segmentation based on language environment in disambiguation and word segmentation efficiency.
-
Review1998, 12(4): 61-66.Based on the analysis of machine codes for Chinese characters ,this paper proposed a dynamic hashing algorithm for quick search of Chinese characters ,which adopt s simple xor operation to disperse all the probable Chinese characters into 256 groups equally and follows a linear search within each group. Experiment s show that the algorithm is of practical value and it s asymptotic time complexity is O (1) .
