In this paper, we claim that Vector Cosine - which is generally considered one of the most efficient unsupervised measures for identifying word similarity in Vector Space Models - can be outperformed by a completely unsupervised measure that evaluates the extent of the intersection among the most associated contexts of two target words, weighting such intersection according to the rank of the shared contexts in the dependency ranked lists. This claim comes from the hypothesis that similar words do not simply occur in similar contexts, but they share a larger portion of their most relevant contexts compared to other related words. To prove it, we describe and evaluate APSyn, a variant of Average Precision that - independently of the adopted parameters - outperforms the Vector Cosine and the co-occurrence on the ESL and TOEFL test sets. In the best setting, APSyn reaches 0.73 accuracy on the ESL dataset and 0.70 accuracy in the TOEFL dataset, beating therefore the non-English US college applicants (whose average, as reported in the literature, is 64.50%) and several state-of-the-art approaches.

Morpheme inversion is a significant lexical phenomenon in the evolution of Chinese words, and it poses additional difficulties in Chinese word segmentation, especially in computer processing of Chinese classics. This paper reports a study on the disyllabic morpheme-inverted compounds in the Chinese diachronic corpora from the perspective of natural language processing. The corpora include two pre-Qin classics and four notable novels created in the Ming and Qing dynasties, in which words are segmented and proper nouns are tagged. Based on the full statistics and analysis, a comparative study is carried out on the use of disyllabic morpheme-inverted compounds in the two types of Chinese text, that is, historical works and fictions. Results show that there are many more morpheme-inverted compounds in the Ming-Qing novels than in the pre-Qin classics in terms of both lexical item and frequency. The morpheme-inverted compounds in the Ming-Qing novels are also closer to their modern counterparts.

The Chinese language is evolving over the centuries. In order to study the changes of Chinese language using computational methods, segmentation and PoS tagging of Chinese are essential. However, segmentation and PoS tagging methods developed for Modern Standard Chinese do not perform well for Classical Chinese. The cost of segmenting and annotation is high if they are done manually. In this work, we present a CRF based method for PoS tagging for Classical Chinese text in the Ming and Qing dynasties. One of the key issues is the preparation of the training data for CRF. Our initial experiment shows that PoS tagging based on Modern Standard Chinese text can achieve a precision of 83%; and by adding as little as 12,000-word annotated Classical Chinese texts, we were able to improve the precision to over 90%.

The quality of text segmentation and annotation plays a significant role in Natural Language Processing especially in downstream applications. This paper presents the specification for word segmentation and named entity annotation targeted for novels in the Ming and Qing dynasties. The purpose of this work is to build the foundational work for computer-aided lexical semantic analysis of classical Chinese literature, especially the transition of Chinese literature from its traditional forms such as traditional verses and vernacular styles to modern Chinese. To assist in literature study, an elaborate named entity annotation scheme is specially developed for classical Chinese. Computer-aided segmentation and named entity annotation are conducted on some famous Ming and Qing Chinese classics. The specification for the segmentation and annotation is produced based on the studies of the morphology and semantics differences as well as similarities between classical Chinese and modern Chinese with reference to the existing standards for modern Chinese processing widely used in Mainland China and Taiwan.

We adopt the corpus-informed approach to example sentence selections for the construction of a reference grammar. In the process, a database containing sentences that are carefully selected by linguistic experts including the full range of linguistic facts covered in an authoritative Chinese Reference Grammar is constructed and structured according to the reference grammar. A search engine system is developed to facilitate the process of finding the most typical examples the users need to study a linguistic problem or prove their hypotheses. The database can also be used as a training corpus by computational linguists to train models for Chinese word segmentation, POS tagging and sentence parsing.