The impact of deep learning on document classification using semantically rich representations

This paper presents a semantically rich document representation model for automatically classifying financial documents into predefined categories utilizing deep learning. The model architecture consists of two main modules including document representation and document classification. In the first module, a document is enriched with semantics using background knowledge provided by an ontology and through the acquisition of its relevant terminology. Acquisition of terminology integrated to the ontology extends the capabilities of semantically rich document representations with an in depth-coverage of concepts, thereby capturing the whole conceptualization involved in documents. Semantically rich representations obtained from the first module will serve as input to the document classification module which aims at finding the most appropriate category for that document through deep learning. Three different deep learning networks each belonging to a different category of machine learning techniques for ontological document classification using a real-life ontology are used.Multiple simulations are carried out with various deep neural networks configurations, and our findings reveal that a three hidden layer feedforward network with 1024 neurons obtain the highest document classification performance on the INFUSE dataset. The performance in terms of F1 score is further increased by almost five percentage points to 78.10% for the same network configuration when the relevant terminology integrated to the ontology is applied to enrich document representation. Furthermore, we conducted a comparative performance evaluation using various state-of-the-art document representation approaches and classification techniques including shallow and conventional machine learning classifiers.     

A systematic review of visual representations for analyzing collaborative discourse

Visual analytics combines automated data analysis and human intelligence through visualisation techniques to address the complexity of current real-world problems. This review uses the lens of visual analytics to examine four dimensions of visual representations for analysing collaborative discourse: goals, data sources, visualisation designs, and analytical techniques based on 89 studies. We found visual analysis approaches to be suitable and advantageous for decomposing the temporality of collaborative discourse. However, it has been challenging for current research to simultaneously consider learning theories and follow visualisation design principles when adopting visualisations to analyse collaborative discourse. At the same time, existing visual analysis approaches have mainly targeted learners or researchers in online contexts and mainly focused on mirroring collaborative discourse rather than providing advanced affordances such as alerting or advising. Informed by these findings, we propose a possible future research agenda and offer suggestions for the features of successful collaboration to guide the design of advanced affordances.     

论复元素(α βi,γ δi)的KM表示法

本文经过分析论证,给出复元(α βi,γ δi)的KM表示法,据此就可将实平面拓广到复平面,在复平面上就可分别作出所有函数y=f(x)的完整的几何图象,并举例阐述其具体应用。     

递归知识的表示及其推理方法

知识表示是研究工程规范管理系统首先要解决的问题，笔者提出了用状态知识元与决策表相结合的方法来表示递归知识，并介绍了相应的推理方法。     

基于行为-内容融合模型的用户画像研究

[目的/意义]为识别并去除非理性投资者的网络评论,提升评论的专业程度与质量,促进理性投资,本文以识别股吧中的用户是否属于噪声投资者为研究任务,进行用户画像。[方法/过程]对股吧的用户发文内容进行深度用户表示学习（deep user representation learning）,结合股吧用户的粉丝数量、影响力、关注量、自选股、吧龄、发帖量、评论量、访问量等行为特征,提出一种行为-内容融合模型（behaviour and content combined model,BCCM）,并在标注数据集上进行实证与对比研究。[结果/结论]实验结果显示,该模型对噪声投资者识别的F1值为79.47%,优于决策树方法（69.90%）、SVM方法（75.61%）、KNN方法（73.21%）和ANN方法（74.83%）。在噪声投资者识别这一特定用户画像研究任务中,通过利用深度用户表示学习引入文本内容特征,能够显著提升用户画像的各种评价指标。     

中学化学中的离子反应与高考中的离子反应方程式

This thesis discusses the judgment of the ionic reaction and ionic reaction representation.     

关于单位分数问题的一些新结论

用初等的方法研究了某些分数的第一类好表法与奇数强表法,得出了一些有用的新结论,并回答了文献[5]中由RonHardin和NeilSoane提出的一个问题.     

How Ordinary is the Ordinary (Wo)man on the Street?

Interviews with the ordinary man or woman on the street are omnipresent in television news. These vox pop interviews are used to represent the general public in the news. Several editorial and practical guidelines exist about the characteristics of a “good” vox pop. But what characteristics do journalists search for in vox pops in practice? This study answers this question by looking at visual and contextual characteristics of vox pop interviews as a means to gain a better understanding of which vox pops appear in the news as a representation of the entire population. We conducted a content analysis of 2000 vox pop interviews in Flanders, Belgium’s Dutch-speaking region, supplemented by interviews with television journalists. We find that, despite editorial guidelines to do so, journalists hardly ever contextualize vox pop interviews by clarifying that they are not necessarily a good representation of the entire population. The results show that journalists select vox pops which are representative of age and gender, but not of minority groups such as ethnic-cultural minorities and people with disabilities. In some regards, vox pops thus provide a biased representation of the population and might influence the public to make wrongful generalizations about public opinion.     

初中思想政治课课堂教学方法设计之我见

初中思想政治课教师在课堂教学方法的设计中，应从实际出发，不断创新，使思想政治课教学始终充满活力。     

关于数学概念表征层次的研究

对于同一个数学概念,不同学生的表征方式是多种多样的.数学概念的表征层次具有某种纵向上的发展性倾向.学生对数学概念的表征并不一定是概念的定义,而可能是与概念定义同构或拟同构的表象,也可能是对概念的自我“修正”.概念表征的方式具有多样性,反映学生对概念的理解水平.数学概念表征网络中,对于概念的表象多以“标准图形”、“原型”、“特殊事例”为主.这些表象对掌握数学概念的本质起到过有益的作用,但对后继学习与运用也有极大的干扰作用.