TY - JOUR
T1 - Common Sense Knowledge for Handwritten Chinese Text Recognition
AU - Wang, Qiu Feng
AU - Cambria, Erik
AU - Liu, Cheng Lin
AU - Hussain, Amir
N1 - Funding Information:
Acknowledgments This work has been supported in part by the National Basic Research Program of China (973 Program) Grant 2012CB316302, the National Natural Science Foundation of China (NSFC) Grants 60825301 and 60933010, and the Royal Society of Edinburgh (UK) and the Chinese Academy of Sciences within the China-Scotland SIPRA (Signal Image Processing Research Academy) Programme. The authors would like to thank Jia-jun Zhang for his aid in the machine translation process.
PY - 2013/6
Y1 - 2013/6
N2 - Compared to human intelligence, computers are far short of common sense knowledge which people normally acquire during the formative years of their lives. This paper investigates the effects of employing common sense knowledge as a new linguistic context in handwritten Chinese text recognition. Three methods are introduced to supplement the standard n-gram language model: embedding model, direct model, and an ensemble of these two. The embedding model uses semantic similarities from common sense knowledge to make the n-gram probabilities estimation more reliable, especially for the unseen n-grams in the training text corpus. The direct model, in turn, considers the linguistic context of the whole document to make up for the short context limit of the n-gram model. The three models are evaluated on a large unconstrained handwriting database, CASIA-HWDB, and the results show that the adoption of common sense knowledge yields improvements in recognition performance, despite the reduced concept list hereby employed.
AB - Compared to human intelligence, computers are far short of common sense knowledge which people normally acquire during the formative years of their lives. This paper investigates the effects of employing common sense knowledge as a new linguistic context in handwritten Chinese text recognition. Three methods are introduced to supplement the standard n-gram language model: embedding model, direct model, and an ensemble of these two. The embedding model uses semantic similarities from common sense knowledge to make the n-gram probabilities estimation more reliable, especially for the unseen n-grams in the training text corpus. The direct model, in turn, considers the linguistic context of the whole document to make up for the short context limit of the n-gram model. The three models are evaluated on a large unconstrained handwriting database, CASIA-HWDB, and the results show that the adoption of common sense knowledge yields improvements in recognition performance, despite the reduced concept list hereby employed.
KW - Common sense knowledge
KW - Handwritten Chinese text recognition
KW - Linguistic context
KW - Natural language processing
KW - n-gram
UR - http://www.scopus.com/inward/record.url?scp=84877086482&partnerID=8YFLogxK
U2 - 10.1007/s12559-012-9183-y
DO - 10.1007/s12559-012-9183-y
M3 - Article
AN - SCOPUS:84877086482
SN - 1866-9956
VL - 5
SP - 234
EP - 242
JO - Cognitive Computation
JF - Cognitive Computation
IS - 2
ER -