Abstract
More and more linked open data and user generated contents are published on the Webafter the Semantic Web has been proposed; the web is now changing from the web ofdocument to web of data, which contains abundant entities and relations. Knowledge graphhas been first proposed by Google,which is focused on describing various real world entitiesand concepts, and their relations. Knowledge graph is a new vision of ontology; a knowledgegraph extends an ontology in the entity level. Ontology usually focuses on concepts and theirrelations, which specifies the schema of knowledge graph; and knowledge graph adds a largenumber of entities into ontology. Knowledge graph is widely used in semantic search,intelligent question-answering, knowledge engineering, data mining, and digital library.
This thesis studies knowledge construction from multiple data sources based on thecurrent research achievements about knowledge graph and ontology construction. The mainworks and contributions of this thesis are as follows,1.This thesis exploits the structured and semi-structured data on the web for knowledgeconstruction, such as linked open data, online encyclopedias, and domain web sites, whichhave high coverage and quick update speed. This thesis explains how to extract and learnknowledge from them, and how to ensure the quality of the constructed knowledge graph.
2.We study on construct knowledge graph from multiple data sources and combine theadvantages of them, including the good precision of the data in relation databases, the highcoverage of the linked open data and public knowledge bases on the web, and the depth of thedomain oriented data. The thesis proposes a knowledge construction method from thesemultiple resources, and ensures the precision of the learnt knowledge graph based on theredundancies of different resources.
3.This thesis also explains how to exact knowledge from large-scale web text andproposes an open relation extraction method based on self-supervised learning. The extractedrelations include synonym, hyponymy and attribute relations among concepts and entities.
The main advantage of the method is that it labels training samples automatically by usingknowledge extracted from structured data or semi-structured data and some general heuristicrules. In order to obtain text automatically, this thesis also proposes a heuristic rule based webinformation extraction algorithm to extract main content of web pages.
4.For domain knowledge graph construction, this thesis also focuses on how to use thedomain structured data and designs a mapping language which specifies how to map data inrelation databases into knowledge in knowledge graph. This thesis also studies how toautomatically discover domain data source such as open domain knowledge bases andwebsites on the web5 and proposed a corresponding algorithm.
5. This thesis also develops an online collaborative knowledge graph edit platform whichtries to leverage crowd-wisdom for knowledge graph generation. The main advantages of theplatform are its ability for concurrent editing,and it could combine with the automaticallearning algorithms.
Finally, we construct a general knowledge graph with 7,392,384 entities and 60,842,064facts based on the proposed algorithms. While comparing the constructed knowledge graphwith other knowledge databases and data sets,we find it has good coverage; meanwhile, theaverage precision of the knowledge is above 95%. Moreover, we construct a domainknowledge graph about fishes which contains more than 32 thousands fish species; its goodcoverage comes from the usage of existing most complete data sources.
Abstract
The present paper explores the categorization of Chinese and English three-wordidioms and discusses the differences of the syntactic functions between those idioms of thetwo languages. The categorization of three-word idioms is motivated by the classification ofidioms in Fillmore et al.⑴ and the autonomy/dependence alignment [2]. To improve theclassification, we also borrow the concept of chemical-bond. In the present study, verbs in thethree-word idioms are seen as equal to the metallic ions in chemistry. Both Chinese andEnglish three-word idioms can be divided into ionic-bond type and covalent-bond type.
Moreover, based on the model of the autonomy/dependence alignment, the Chinesethree-word idioms can be further classified into 11 subtypes, while English three-word idiomswill be categorized into 9 subtypes.
The contrastive research of the syntactic functions is conducted on the basis of theclassification. We start with the contrastive study of the categorization between Chinese andEnglish three-word idioms. Qualitative study is the first step in the discussion of the syntacticfunctions. The examples are obtained through the corpus and the syntactic functions of thethree-word idioms will be analyzed within the sentences. Indeed, we can draw someconclusions from the qualitative study, but the lack of representativeness and reliability makesthe study incomplete, calling forth the quantitative study. In the present paper, 1000 Chinesethree-word idioms and 1000 English three-word idioms are picked up from the Dictionary ofIdioms from the Ancient Chinese Novels and NTC 's American Idioms Dictionary respectively.
With the assistance of SPSS and corpus, we have done some data processing on the syntacticfunctions of the three-word idioms. A special note for this is that the chi-square test in SPSSwill prove the reliability and representativeness of the sample test in the present study.
We find, first of all, 32.3% of the English covalent-bond idioms act as objects orsubjects in the sentences and 67.7% as modifiers or complements; 25.7% of the Chinesecovalent-bond idioms act as objects or subjects and 74.3% as modifiers or complements.
Secondly, all of the English ionic-bond idioms can function as predicates and 97.8% of themcan also be modifiers or complements; 86.5% of Chinese ionic-bond idioms act as predicatesand 13.5% as modifiers or complements. According to this, the difference and similarity ofthe syntactic functions between Chinese and English three-word idioms can be concluded asfollows: all the English ionic-bond idioms act as predicates while in Chinese over 80% of theionic-bond idioms function as predicates leaving 13.5% functioning as modifiers orcomplements; the syntactic functions of covalent-bond idioms are similar in Chinese andEnglish: about 70% of the covalent-bond idioms function as modifiers or complements whileabout 30% act as objects or subjects.
The findings of this contrastive study will help analyze the syntactic functions of thethree-word idioms in the sentences and assist the translation of the three-word idioms.
Moreover, the present study is the first research on the classification of three-word idioms,which, we hope, will be of help for the study of idioms.
Abstract
Golf apparel is categorized as outdoor wears. Originated in the 14th centuryin Scotland, golf was an aristocratic sport with a long history and it was voted asthe competition event in the 2016 Olympic Games. As a sport inheriting thearistocratic tradition and culture, golf apparel also underwent drastic reforms andevolutions from a traditional aristocratic phase to a modern style and from acomplicated pattern to a simple trend. Golf apparel is categorized as outdoorclothing catalogue in the western part of the dress system. Unlike athleticsportswear, it likes casual clothing rather than athletic sportswear. Like hunting,shooting, horse racing and fishing which English-speaking country gentlemenpursued, golf represents a way of life. Golf apparel experienced a change fromcomplex to simple and saw the creation of some classic outdoor clothing such asthe Norfolk jackets, casual jacket, Polo-shirt, etc. This manuscript focused onanalyzing the evolution of golf apparel from the 20th century till today. Historicalorigins and perspective of the golf apparel was discussed to facilitate theunderstanding of its history and culture. This study also investigated therevolution of the golf apparel inspired by the modern technology and in reverse,its long-term impact on the daily lifestyles.
Abstract
With the rise of international communication, learning English has become anindispensible task faced by the minority nationalities in China. Hence, the number oftrilingual is rocketing. Among them, Mongolian trilingual is featured with uniqueness.
This article involves four groups of cognitive experiment from visual and auditorychannels, aiming to explore the time course of semantic activation in the process oflexical access, the impact of characters and speech on lexical access of English andthe relationship of L3 （English） to LI （Mongolian） and L3 （English） to L2 （Chinese）。
The design is based on event-related potentials （ERP technology）， which records thesensory activity and cognitive process online continuously. Separating ERPcomponents from the scalp to determine the psychological activity of activated brainarea, we cm locate the endogenous nerve of overt behavior. The study results serve asan effective supplement in cognitive study of vocabulary. Besides,it can providetheoretical guidance for English teaching.
Specific conclusions include:1）The exits of priming effect in all four groups of experiment indicate the lexicalconcept of Mongolian, Chinese and English is stored together. All three languages hasshared semantic layer but different concept layers, and the conceptual representationof English （L3） is accessed through translation of Mongolian （LI） and Chinese （L2）as mediation;2）In terms of the interaction of three languages, Mongolian college students arebetter at the judgment task when the material presented is L3 （English） to L2 （Chinese）rather than L3 （English） to LI （Mongolian），which can be deduced that the thirdlanguage acquisition （English） depends largely on the second language （Chinese）rather than their mother tongue （Mongolian）；3）The emergence of P300 indicates the processing of speech take place beforesemantic processing of English vocabulary. In addition, the semantic processing takesshorter time though the representation of speech than characters. Those facts proveMongolian trilingual achieve semantic access with the help of speech activation,which is largely affected by their mother tongue.
Abstract
Emotional meaning plays an important role in discourse comprehension. It is oftheoretical significance and important application value to deeply .reveal readers'specific cognitive processes in mind when they comprehend emotions in discourse.
However, traditional literary analysis regards emotions in discourse as rhetorical effectsand results of artistic expressions with internal comprehension processes neglected.
Psychological linguistics and computational linguistics treat them as directinterpretation or reasoning, equating propositional information with comprehensionresults. Cognitive poetics reveals that it is the tangible emotional responses that aretriggered by emotional comprehension in discourse, which will adversely affect thecomprehension processing of discourse. In this process, readers' self-concept andemotional experience play a vital part. But what previous studies lack is the specificdiscussion of internal processes on these phenomena.
Mental spaces and conceptual integration theory （MSCI for short） target describingmid explaining online meaning construction. Recently, researches on poetry,novels andsuper-long texts by means of MSCI are constantly emerging, which have been makinglots of contributions to discourse interpretation. However, as Freeman （2006） pointedout that describing and explaining emotions in discourse is still in the blank. Thisdissertation attempts to solve two problems by means of MSCI: the first one is theonline process involved in the comprehension of emotional meaning; the second one isthe impact of these processes on mental coherence in discourse.
To bridge the gap, this study defines emotion as an episode on the basis of therecent development of cognitive science,and analyzes the basic departments and mentalcomponents of emotional process and the relationships among these ingredientsaccording to the episode model of emotions. Then, the on-line emotional meaningconstruction in discourse comprehension is analyzed with examples, whichdistinguishes different emotional mental spaces according to their mental mechanism.
Particularly, integration processes involved in on-line emotional meaning constructionin discourse comprehension are probed, including the nature of cognitive operation andvital relations such as selective mappings, etc. Furthermore, an experiment is conductedto investigate the understanding differences under the condition of different emotionalprompts. Finally we analyze the connecting and nested relation between emotionalmental spaces formed through on-line emotional meaning construction in discoursecomprehension and the corresponding conceptual integration process with examples.
More than that, the differences are compared between mental coherence in discoursetaken shape by emotional mental spaces and their conceptual integration processes andthe structure formed by traditional discrete factors like time, space, cause-effect. What'smore, this paper attempts to analyze medium and macroscopic structure of discoursefrom the perspective of emotions.
The findings of this research are concluded as the following: Firstly, thecomprehension process must contain conceptual integration process of activatinglong-term memory to build up experiential space, and the fusion of experiential spaceand narrative space to shape into blended space. Blended space demonstrates theemergent property of emotional mental space in the case of emotional arousals excludedand included. “Event cognition-emotion” relation, “emotion-consequent behavior”
relation and “emotion-cognitive color” relation are the very vital relations in thecomprehension process of emotions. These vital relations are the cognitive mechanismof “emotion-cognition” interaction. “Emotion-cognition” interactive mechanism isthe specific mechanism of cognitive operation （including composition, completion,elaboration and so on） in emotional mental space. It makes emotional and narrativedescription in discourse mutually nurish and promote to manifest the gestalt effect （Thewhole is greater than the sum of its parts.）， engendering rich beyond literal meaningsand two-way value-added pragmatic effect.
Secondly, It is demonstrated that “event cognitionemotion” relation,“emotion~consequent behavior” relation,“emotion-~cognitive color” relation andemotional contrast structure make emotional information in discourse （often constructedbeyond literally） and rich narrative content naturally joint together, and unite characters,scenes, events, etc., in related contents. Emotion executes functions of reference,consistency, governace, reason, and integration in these joints. On the macroscopic level,emotions activated in online discourse processing by the reader can implementfunctions of compression,generalization, mark and layout, etc. for information indiscourse to form global thematic impression of discourse and outline structure. Thecoherent relationship and structure ground on emotional mental space system is a newway to take shape coherent discourse and structure different from general cognitivefactors, such as time, space, logic, comparison, analogy. It manifests the dominance ofemotional processing and special gestalt mental representation in discourse by“emotion-cognition” interactive mechanism.
In general, this research explains specific emotional meaning by the use of theintegrated relationship between emotional arousals and cognitive processing to build upthe compositional view of emotional comprehension. It classifies “eventcognitionemotion” relation, emotion一consequent behavior,' relation,“emotion~cognitive color'' relation and emotional contrast structure on the strength ofthe analysis of emotional process to form a unified theoretical framework, based onwhich we analyze on-line meaning construction and coherent structure of emotionaldiscourse. It also points out the necessity of conceptual integration in the process ofemotional understanding in discourse,which makes emotional understanding embodiedand achieves gestalt comprehension effect through interaction between cognition andemotion. What's more, it proposes a new perspective of discourse coherence, namelyemotional mental spaces, to analyze local, medium and macroscopic structure ofcoherence, and then discovers the independent structural system of implicity, depth,high order and integration in emotional discourse. Accordingly, the above innovativework in this study extends the application of MSCI and enriches the explanation ofCognitive Linguistics to discourse structure and coherence. What's more, this study alsoattempts to testify MSCI with empirical evidences.
The main limitation of this study is focusing on static discourse. In addition, theanalyzing materials are confined to Chinese and lack multilingual contrast andcomparison. Future research should further discuss interactive discourse analysis andcontrast and comparison of cross-cultural communication.
