A two-stage approach for generating topic models

Topic modeling has been widely utilized in the fields of information retrieval, text mining, text classification etc. Most existing statistical topic modeling methods such as LDA and pLSA generate a term based representation to represent a topic by selecting single words from multinomial word distribution over this topic. There are two main shortcomings: firstly, popular or common words occur very often across different topics that bring ambiguity to understand topics; secondly, single words lack coherent semantic meaning to accurately represent topics. In order to overcome these problems, in this paper, we propose a two-stage model that combines text mining and pattern mining with statistical modeling to generate more discriminative and semantic rich topic representations. Experiments show that the optimized topic representations generated by the proposed methods outperform the typical statistical topic modeling method LDA in terms of accuracy and certainty.

Experiential knowledge representation and the design of product usability

The topic of designers’ knowledge and how they conduct design process has been widely investigated in design research. Understanding theoretical and experiential knowledge in design has involved recognition of the importance of designers’ experience of experiencing, seeing, and absorbing ideas from the world as points of reference (or precedents) that are consulted whenever a design problem arises (Lawson, 2004). Hence, various types of design knowledge have been categorized (Lawson, 2004), and the nature of design knowledge continues to be studied (Cross, 2006); nevertheless, the study of the experiential aspects embedded in design knowledge is a topic not fully addressed. In particular there has been little emphasis on the investigation of the ways in which designers’ individual experience influences different types of design tasks. This research focuses on the investigation of the ways in which designers inform a usability design process. It aims to understand how designers design product usability, what informs their process, and the role their individual experience (and episodic knowledge) plays within the design process. This paper introduces initial outcomes from an empirical study involving observation of a design task that emphasized usability issues. It discusses the experiential knowledge observed in the visual representations (sketches) produced by designers as part of the design tasks. Through the use of visuals as means to represent experiential knowledge, this paper presents initial research outcomes to demonstrate how designers’ individual experience is integrated into design tasks and communicated within the design process. Initial outcomes demonstrate the influence of designers’ experience in the design of product usability. It is expected that outcomes will help identify the causal relationships between experience, context of use, and product usability, which will contribute to enhance our understanding about the design of user-product interactions.

A distributed knowledge management infrastructure based on a Topic Map grid

Modern enterprise knowledge management systems typically require distributed approaches and the integration of numerous heterogeneous sources of information. A powerful foundation for these tasks can be Topic Maps, which not only provide a semantic net-like knowledge representation means and the possibility to use ontologies for modelling knowledge structures, but also offer concepts to link these knowledge structures with unstructured data stored in files, external documents etc. In this paper, we present the architecture and prototypical implementation of a Topic Map application infrastructure, the ‘Topic Grid’, which enables transparent, node-spanning access to different Topic Maps distributed in a network.

Pattern-based topic models for information filtering

Topic modelling, such as Latent Dirichlet Allocation (LDA), was proposed to generate statistical models to represent multiple topics in a collection of documents, which has been widely utilized in the fields of machine learning and information retrieval, etc. But its effectiveness in information filtering is rarely known. Patterns are always thought to be more representative than single terms for representing documents. In this paper, a novel information filtering model, Pattern-based Topic Model(PBTM) , is proposed to represent the text documents not only using the topic distributions at general level but also using semantic pattern representations at detailed specific level, both of which contribute to the accurate document representation and document relevance ranking. Extensive experiments are conducted to evaluate the effectiveness of PBTM by using the TREC data collection Reuters Corpus Volume 1. The results show that the proposed model achieves outstanding performance.

Topic dependent language modelling for spoken term detection

This paper investigates the effect of topic dependent language models (TDLM) on phonetic spoken term detection (STD) using dynamic match lattice spotting (DMLS). Phonetic STD consists of two steps: indexing and search. The accuracy of indexing audio segments into phone sequences using phone recognition methods directly affects the accuracy of the final STD system. If the topic of a document in known, recognizing the spoken words and indexing them to an intermediate representation is an easier task and consequently, detecting a search word in it will be more accurate and robust. In this paper, we propose the use of TDLMs in the indexing stage to improve the accuracy of STD in situations where the topic of the audio document is known in advance. It is shown that using TDLMs instead of the traditional general language model (GLM) improves STD performance according to figure of merit (FOM) criteria.

Supervised latent dirichlet allocation models for efficient activity representation

Local spatio-temporal features with a Bag-of-visual words model is a popular approach used in human action recognition. Bag-of-features methods suffer from several challenges such as extracting appropriate appearance and motion features from videos, converting extracted features appropriate for classification and designing a suitable classification framework. In this paper we address the problem of efficiently representing the extracted features for classification to improve the overall performance. We introduce two generative supervised topic models, maximum entropy discrimination LDA (MedLDA) and class- specific simplex LDA (css-LDA), to encode the raw features suitable for discriminative SVM based classification. Unsupervised LDA models disconnect topic discovery from the classification task, hence yield poor results compared to the baseline Bag-of-words framework. On the other hand supervised LDA techniques learn the topic structure by considering the class labels and improve the recognition accuracy significantly. MedLDA maximizes likelihood and within class margins using max-margin techniques and yields a sparse highly discriminative topic structure; while in css-LDA separate class specific topics are learned instead of common set of topics across the entire dataset. In our representation first topics are learned and then each video is represented as a topic proportion vector, i.e. it can be comparable to a histogram of topics. Finally SVM classification is done on the learned topic proportion vector. We demonstrate the efficiency of the above two representation techniques through the experiments carried out in two popular datasets. Experimental results demonstrate significantly improved performance compared to the baseline Bag-of-features framework which uses kmeans to construct histogram of words from the feature vectors.

Efficient classification using phrases generated by topic models

There are many popular models available for classification of documents like Naïve Bayes Classifier, k-Nearest Neighbors and Support Vector Machine. In all these cases, the representation is based on the “Bag of words” model. This model doesn't capture the actual semantic meaning of a word in a particular document. Semantics are better captured by proximity of words and their occurrence in the document. We propose a new “Bag of Phrases” model to capture this discriminative power of phrases for text classification. We present a novel algorithm to extract phrases from the corpus using the well known topic model, Latent Dirichlet Allocation(LDA), and to integrate them in vector space model for classification. Experiments show a better performance of classifiers with the new Bag of Phrases model against related representation models.

Standardizing clinical trials workflow representation in UML for international site comparison.

BACKGROUND: With the globalization of clinical trials, a growing emphasis has been placed on the standardization of the workflow in order to ensure the reproducibility and reliability of the overall trial. Despite the importance of workflow evaluation, to our knowledge no previous studies have attempted to adapt existing modeling languages to standardize the representation of clinical trials. Unified Modeling Language (UML) is a computational language that can be used to model operational workflow, and a UML profile can be developed to standardize UML models within a given domain. This paper's objective is to develop a UML profile to extend the UML Activity Diagram schema into the clinical trials domain, defining a standard representation for clinical trial workflow diagrams in UML. METHODS: Two Brazilian clinical trial sites in rheumatology and oncology were examined to model their workflow and collect time-motion data. UML modeling was conducted in Eclipse, and a UML profile was developed to incorporate information used in discrete event simulation software. RESULTS: Ethnographic observation revealed bottlenecks in workflow: these included tasks requiring full commitment of CRCs, transferring notes from paper to computers, deviations from standard operating procedures, and conflicts between different IT systems. Time-motion analysis revealed that nurses' activities took up the most time in the workflow and contained a high frequency of shorter duration activities. Administrative assistants performed more activities near the beginning and end of the workflow. Overall, clinical trial tasks had a greater frequency than clinic routines or other general activities. CONCLUSIONS: This paper describes a method for modeling clinical trial workflow in UML and standardizing these workflow diagrams through a UML profile. In the increasingly global environment of clinical trials, the standardization of workflow modeling is a necessary precursor to conducting a comparative analysis of international clinical trials workflows.

Crisis of political representation and democratization in Mexico: the generality and specificity of the case

This article attempts to assess the implications and the own character of the crisis of representation in Mexico. Once the topic framed and the long-term dynamics of Mexican political elites presented, this paper will attempt to understand why, despite the pluralization of the party system, there remain many questions about the truly democratic nature of the Mexican political system.

Employees' external representation of their workplace: key antecedents

Drawing on attitude theory, this study investigates the drivers of employees' expression of favorable opinions about their workplace. Despite its theoretical and managerial importance, the marketing literature largely ignores the topic. This study advances prior research by developing, and empirically testing, a conceptual framework of the relationship between workgroup support and favorable external representation of the workplace, mediated by emotional responses to this support. The present research investigates four new relationships: between workgroup support and emotional exhaustion, workgroup support and organizational commitment, workgroup support and job satisfaction, and emotional exhaustion and external representation of the workplace. Based on a sample of over 700 frontline service employees, this study finds that workgroup support affects favorable external representation of the workplace through various emotional responses (i.e., emotional exhaustion, organizational commitment and job satisfaction). In addition, the results identify employees' organizational commitment as the most important determinant of favorable external representation of the workplace, followed by job satisfaction and reduced emotional exhaustion. These results suggest that companies should develop practices that encourage workgroup support and organizational commitment to achieve favorable external representation of the workplace.

Differences in the perception of interest representation: a comparision of Brasília and Brussels lobbying activity

The research topic of this paper is focused on the analysis of how trade associations perceive lobbying in Brussels and in Brasília. The analysis will be centered on business associations located in Brasília and Brussels as the two core centers of decision-making and as an attraction for the lobbying practice. The underlying principles behind the comparison between Brussels and Brasilia are two. Firstof all because the European Union and Brazil have maintained diplomatic relations since 1960. Through these relations they have built up close historical, cultural, economic and political ties. Their bilateral political relations culminated in 2007 with the establishment of a Strategic Partnership (EEAS website,n.d.). Over the years, Brazil has become a key interlocutor for the EU and it is the most important market for the EU in Latin America (European Commission, 2007). Taking into account the relations between EU and Brazil, this research could contribute to the reciprocal knowledge about the perception of lobby in the respective systems and the importance of the non-market strategy when conducting business. Second both EU and Brazilian systems have a multi-level governance structure: 28 Member States in the EU and 26 Member States in Brazil; in both systems there are three main institutions targeted by lobbying practice. The objective is to compare how differences in the institutional environments affect the perception and practice of lobbying, where institutions are defined as ‘‘regulative, normative, and cognitive structures and activities that provide stability and meaning to social behavior’’ (Peng et al., 2009). Brussels, the self-proclaimed "Capital of Europe”, is the headquarters of the European Union and has one of the highest concentrations of political power in the world. Four of the seven Institutions of the European Union are based in Brussels: the European Parliament, the European Council, the Council and the European Commission (EU website, n.d.). As the power of the EU institutions has grown, Brussels has become a magnet for lobbyists, with the latest estimates ranging from between 15,000 and 30,000 professionals representing companies, industry sectors, farmers, civil society groups, unions etc. (Burson Marsteller, 2013). Brasília is the capital of Brazil and the seat of government of the Federal District and the three branches of the federal government of Brazilian legislative, executive and judiciary. The 4 city also hosts 124 foreign embassies. The presence of the formal representations of companies and trade associations in Brasília is very limited, but the governmental interests remain there and the professionals dealing with government affairs commute there. In the European Union, Brussels has established a Transparency Register that allows the interactions between the European institutions and citizen’s associations, NGOs, businesses, trade and professional organizations, trade unions and think tanks. The register provides citizens with a direct and single access to information about who is engaged in This process is important for the quality of democracy, and for its capacity to deliver adequate policies, matching activities aimed at influencing the EU decision-making process, which interests are being pursued and what level of resources are invested in these activities (Celgene, n.d). It offers a single code of conduct, binding all organizations and self-employed individuals who accept to “play by the rules” in full respect of ethical principles (EC website, n.d). A complaints and sanctions mechanism ensures the enforcement of the rules and addresses suspected breaches of the code. In Brazil, there is no specific legislation regulating lobbying. The National Congress is currently discussing dozens of bills that address regulation of lobbying and the action of interest groups (De Aragão, 2012), but none of them has been enacted for the moment. This work will focus on class lobbying (Oliveira, 2004), which refers to the performance of the federation of national labour or industrial unions, like CNI (National Industry Confederation) in Brazil and the European Banking Federation (EBF) in Brussels. Their performance aims to influence the Executive and Legislative branches in order to defend the interests of their affiliates. When representing unions and federations, class entities cover a wide range of different and, more often than not, conflicting interests. That is why they are limited to defending the consensual and majority interest of their affiliates (Oliveira, 2004). The basic assumption of this work is that institutions matter (Peng et al, 2009) and that the trade associations and their affiliates, when doing business, have to take into account the institutional and regulatory framework where they do business.

Knowledge Representation and Orthophemism: A Reflection Aiming to a Concept

Characteristics of speech, especially figures of speech, are used by specific communities or domains, and, in this way, reflect their identities through their choice of vocabulary. This topic should be an object of study in the context of knowledge representation once it deals with different contexts of production of documents. This study aims to explore the dimensions of the concepts of euphemism, dysphemism, and orthophemism, focusing on the latter with the goal of extracting a concept which can be included in discussions about subject analysis and indexing. Euphemism is used as an alternative to a non-preferred expression or as an alternative to an offensive attribution-to avoid potential offense taken by the listener or by other persons, for instance, pass away. Dysphemism, on the other hand, is used by speakers to talk about people and things that frustrate and annoy them-their choice of language indicates disapproval and the topic is therefore denigrated, humiliated, or degraded, for instance, kick the bucket. While euphemism tries to make something sound better, dysphemism tries to make something sound worse. Orthophemism (Allan and Burridge 2006) is also used as an alternative to expressions, but it is a preferred, formal, and direct language of expression when representing an object or a situation, for instance, die. This paper suggests that the comprehension and use of such concepts could support the following issues: possible contributions from linguistics and terminology to subject analysis as demonstrated by Talamo et al. (1992); decrease of polysemy and ambiguity of terms used to represent certain topics of documents; and construction and evaluation of indexing languages. The concept of orthophemism can also serves to support associative relationships in the context of subject analysis, indexing, and even information retrieval related to more specific requests.

Warts and All: The Representation and Interpretation of War in Museums

Thesis (Master's)--University of Washington, 2016-06

Multi-topic information filtering with a single user profile

In Information Filtering (IF) a user may be interested in several topics in parallel. But IF systems have been built on representational models derived from Information Retrieval and Text Categorization, which assume independence between terms. The linearity of these models results in user profiles that can only represent one topic of interest. We present a methodology that takes into account term dependencies to construct a single profile representation for multiple topics, in the form of a hierarchical term network. We also introduce a series of non-linear functions for evaluating documents against the profile. Initial experiments produced positive results.

Designing a safety management support system using open architecture databases: the requirements of information representation in the support of managers' risk-related decision-making

This research was conducted at the Space Research and Technology Centre o the European Space Agency at Noordvijk in the Netherlands. ESA is an international organisation that brings together a range of scientists, engineers and managers from 14 European member states. The motivation for the work was to enable decision-makers, in a culturally and technologically diverse organisation, to share information for the purpose of making decisions that are well informed about the risk-related aspects of the situations they seek to address. The research examined the use of decision support system DSS) technology to facilitate decision-making of this type. This involved identifying the technology available and its application to risk management. Decision-making is a complex activity that does not lend itself to exact measurement or precise understanding at a detailed level. In view of this, a prototype DSS was developed through which to understand the practical issues to be accommodated and to evaluate alternative approaches to supporting decision-making of this type. The problem of measuring the effect upon the quality of decisions has been approached through expert evaluation of the software developed. The practical orientation of this work was informed by a review of the relevant literature in decision-making, risk management, decision support and information technology. Communication and information technology unite the major the,es of this work. This allows correlation of the interests of the research with European public policy. The principles of communication were also considered in the topic of information visualisation - this emerging technology exploits flexible modes of human computer interaction (HCI) to improve the cognition of complex data. Risk management is itself an area characterised by complexity and risk visualisation is advocated for application in this field of endeavour. The thesis provides recommendations for future work in the fields of decision=making, DSS technology and risk management.