Pengyuan Liu

Hedge Detection with Latent Features

In the current internet environment, the issue of information credibility is crucial given that anyone can put up information on the Web freely. Hence, it is urge and necessary to develop computational approaches to distinguish reliable information from the unreliable, inaccurate one. In the year 2010, CoNLL proposed a new shared task of hedge detection, aiming to push un-certainty detection for natural language processing applications. Among the pro-posed approaches, sequence labeling models exhibit promising performance. However, only shallow features (e.g. word, lemma, POS tags, etc.) were explored in the existing research. In this paper, we aim to exploit the advantage of topical features in sequence labeling based hedge detection. The experimental results illustrate the effectiveness of the high-level features.

A Hypothesis on Word Similarity and Its Application

A hypothesis is proposed: the semantic distance between the synonyms or near-synonyms should have the same characteristic as the distance in a metrics space. Metrics space is a set where a notion of distance (called a metric) between elements of the set is defined. At the same time, three properties should be held: (i) Identity of Indiscernibles – the distance is zero if and only if the two elements are the same. (ii) Symmetry – The distance between element A and B is equal to the distance between element B and A. (iii) Triangle Inequality – Given three elements A, B and C, the sum of any two pairs’ distance is greater or equal to the rest one. The first two properties is reasonable intuitively; as to the last one, we first get the word similarities based on HowNet and check whether the synonyms or near-synonyms listed in Cilin Extended Edition can satisfy this property. The experiments show that more than 98.5% triples (consists of three synonyms) satisfy the last property – triangle inequality. Fatherly, we detect a large quantity of thesaurus errors according to our hypothesis.

Quantitative Analysis of Synergetic Properties in Chinese Nouns

This paper analyses more than 13000 nouns in contemporary Chinese and investigates their synergetic properties on the basis of word polysemy. HowNet is used in this paper to measure the word polysemy and it proves to be an effective approach. Statistical analysis of the data indicates that the polysemy of nouns abides by the modified Zipf-Alekseev distribution. The results of function fitting with Altmann Fitter show that word polysemy is somewhat related to some linguistic variables including word length, word frequency and polytexuality, which means that each variables in the subsystem of lexicon is still synergetic and Kohler’s lexical model proves to be effective for Chinese nouns.

Study in Words and Characters Use and Contents of the Opening Addresses by Chinese University Presidents

This paper selects 213 opening addresses by Chinese university presidents between 2003 and 2015, synchronically and diachronically, this paper has devoted itself to the study of the usage of words and characters in those addresses, the contents as a whole. The location and the time those speeches were delivered have also been taken into consideration. By comparing the words and characters used in those addresses to Lexicon of Common Words in Contemporary Chinese, this paper shows that vocabulary used in addresses doesn’t change dramatically with time going diachronically, and the themes and focuses of the speeches tend to vary from time to time. This paper classifies the speeches into certain types according to the frequently and individually used words, analyzes the features of their contents. The conclusion is that universities have put more emphasis on cultivating spirits, quality, ability and virtue.

A Corpus-based Study on the Structure of V in “进行[jin4xing2]+V” in Modern Chinese

This paper argues for the necessity of study on the structure of “V” in “进行[jin4xing2]+V” based on the analysis of the previous research, and gives a full description and analysis of it based on the large-scale corpus. In the choice of verb of predicate-object structure, only verbs that have external object in semantic are allowed to enter “进行[jin4xing2]+V” structure. In the syllable, only double-syllable or tri-syllable verbs can be the verb object of “进行[jin4xing2]”, and those tri-syllable verbs are mainly composed in additional morphology, with affixes like”化[hua4]、热[re4]、再[zai4]”and so on.

Computer Semantic Identification of the Adverb “Dou”

Semantic identification of adverbs in Modern Chinese is regarded as an exploration and attempt in improving semantic analysis on sentence level. This paper mainly focuses on two problems. Firstly, how to make computer identify the meanings of adverb “Dou”. Secondly, how to make computer identify the semantic orientation of adverb “Dou”. Based on the real large-scale corpus, the authors investigate the meanings of adverb “Dou” which are restricted by various factors and formal features, summarize the formal rules in a systematic way, and build a flow chart that computers are capable of identifying. We hope that this research will contribute to computer understanding and generating in the sentence that contains the adverb “Dou”.

A Corpus-Based Analysis of Syntactic-Semantic Relations between Adjectival Objects and Nouns in Mandarin Chinese

In this study, we discuss the syntactic and semantic relations between adjectival objects of VA verb-object constructions and specific nouns in Mandarin Chinese. Firstly, it shows that VA verb-object constructions can function as predicates and modifiers in sentences. It also demonstrates that as adjectives represent attributes of nouns, the adjectival objects of VA verb-object constructions have attribute-entity semantic relations with specific nouns. According to the syntactic functions and attribute-entity semantic relations, we point out that there are two main kinds of syntactic relations which are subject-predicate and modifier-head relations between adjectival objects and specific nouns. Moreover, we take adjectival objects as metonymic expressions. At last, we argue that adjectives which function as objects have nominal meanings in semantics, while they are not nominalized in syntax.

First Explorations on the Syntactic Distributions of Verb-Object Constructions in Mandarin–V+A Constructions as Examples

This study aims to use a large-scale corpus to explore the syntactic distributions of V+A bound verb-object constructions which consist of single verbs and single adjectival objects. We find that the distributions of V+A constructions in the corpus come in three varieties: 1.without followed by constituents; 2. followed by nominal constituents; 3.followed by non-nominal constituents. We arrive at the following conclusions that impact on the distributions: 1.the combination degree of V and A; 2.the properties of nominal constituents and the effect of V+A collocate with nominal elements. Then we discuss a special syntactical type such as “da-hei”“zhao-xin”. At last, we find and analyze some verb-object constructions which usually cannot exist independently.

A Research into the Multi-category Words of Verb-Noun in Modern Chinese Based on the Quantitative Analysis of the Collocating Classifiers

The paper studies the multi-category words of verb-noun. We use six words with different parts of speech in two dictionaries to discuss the multi-category words of verb-noun. Based on the collocation between the multi-category words of verb-noun and the corresponding classifiers in a large-scale corpus, we describe the verb-noun words phenomenon considering words’ syntactic characteristics and semantic characteristics. The data shows that di4zhen4, jiu4ji4, pei2tong2, shen1qing3 can collocate with classifiers as nouns or verbs and the collocations are distinct. We could see these four as multi-category words of verb-noun because of their evident noun’s and verb’s characteristics. Feng4xian4, ji4zai3 do not have ample collocations with action measure words, neither do with static measure words, but we also consider these two as multi-category words of verb-noun.

Semantic Analysis of “V+死+人” Structure

In this paper, we found that some Examples of language which represent the structure (This paper explores the structure “V+死+人” while “死” and Monosyllable verbs collocate, and combined into verb-complement structure to collocate with “人” phrase structure.) of “V+死+人” (The“V”in this paper just refer to monosyllable verb.) can express the meaning of “the loss of life or death”, which is meaning (a); some can express the meaning of “to reach the pole”, which is meaning (b); some can both express (a) and (b). Based on the large-scale corpus for obtaining language examples, this paper obtained the total of 112 monosyllable verbs which can enter the structure of “V+死+人”. In this paper, we believe that the structure will express (a) if the semantic orientation point to “人”, and will express (b) if the semantic orientation point to “V”; but characteristics of “V” is the fundamental reason causing the structure of different semantics. Specifically, in the structure of “V+死+人” which expresses (a), the semantic features of “V” is [+damage, +strong action, +death]; in the structure of “V+死+人” which expresses (b), the semantic features of “V” is [+psychology, +quantity, +feelings, -death] or [-psychology, +quantity, +feelings, -death]; in the structure of “V+死+人”which can both express (a) and (b), the semantic features of “V” is [+physiological, feelings, +quantity, +death]