Probabilistic Models for Scalable Knowledge Graph Construction

In the past decade, systems that extract information from millions of Internet documents have become commonplace. Knowledge graphs -- structured knowledge bases that describe entities, their attributes and the relationships between them -- are a powerful tool for understanding and organizing this vast amount of information. However, a significant obstacle to knowledge graph construction is the unreliability of the extracted information, due to noise and ambiguity in the underlying data or errors made by the extraction system and the complexity of reasoning about the dependencies between these noisy extractions. My dissertation addresses these challenges by exploiting the interdependencies between facts to improve the quality of the knowledge graph in a scalable framework. I introduce a new approach called knowledge graph identification (KGI), which resolves the entities, attributes and relationships in the knowledge graph by incorporating uncertain extractions from multiple sources, entity co-references, and ontological constraints. I define a probability distribution over possible knowledge graphs and infer the most probable knowledge graph using a combination of probabilistic and logical reasoning. Such probabilistic models are frequently dismissed due to scalability concerns, but my implementation of KGI maintains tractable performance on large problems through the use of hinge-loss Markov random fields, which have a convex inference objective. This allows the inference of large knowledge graphs using 4M facts and 20M ground constraints in 2 hours. To further scale the solution, I develop a distributed approach to the KGI problem which runs in parallel across multiple machines, reducing inference time by 90%. Finally, I extend my model to the streaming setting, where a knowledge graph is continuously updated by incorporating newly extracted facts. I devise a general approach for approximately updating inference in convex probabilistic models, and quantify the approximation error by defining and bounding inference regret for online models. Together, my work retains the attractive features of probabilistic models while providing the scalability necessary for large-scale knowledge graph construction. These models have been applied on a number of real-world knowledge graph projects, including the NELL project at Carnegie Mellon and the Google Knowledge Graph.

Multiraciality Enters the University: Mixed Race Identity and Knowledge Production in Higher Education

“Multiraciality Enters the University: Mixed Race Identity and Knowledge Production in Higher Education,” explores how the category of “mixed race” has underpinned university politics in California, through student organizing, admissions debates, and the development of a new field of study. By treating the concept of privatization as central to both multiraciality and the neoliberal university, this project asks how and in what capacity has the discourses of multiracialism and the growing recognition of mixed race student populations shaped administrative, social, and academic debates at the state’s flagship universities—the University of California at Berkeley and Los Angeles. This project argues that the mixed race population symbolizing so-called “post-racial societies” is fundamentally attached to the concept of self-authorship, which can work to challenge the rights and resources for college students of color. Through a close reading of texts, including archival materials, policy and media debates, and interviews, I assert that the contemporary deployment of mixed race within the US academy represents a particularly post-civil rights development, undergirded by a genealogy of U.S. liberal individualism. This project ultimately reveals the pressing need to rethink ways to disrupt institutionalized racism in the new millennium.

Are iron oxide nanoparticles safe? Current knowledge and future perspectives

Due to their unique physicochemical properties, including superparamagnetism, iron oxide nanoparticles (ION) have a number of interesting applications, especially in the biomedical field, that make them one of the most fascinating nanomaterials. They are used as contrast agents for magnetic resonance imaging, in targeted drug delivery, and for induced hyperthermia cancer treatments. Together with these valuable uses, concerns regarding the onset of unexpected adverse health effects following exposure have been also raised. Nevertheless, despite the numerous ION purposes being explored, currently available information on their potential toxicity is still scarce and controversial data have been reported. Although ION have traditionally been considered as biocompatible - mainly on the basis of viability tests results - influence of nanoparticle surface coating, size, or dose, and of other experimental factors such as treatment time or cell type, has been demonstrated to be important for ION in vitro toxicity manifestation. In vivo studies have shown distribution of ION to different tissues and organs, including brain after passing the blood-brain barrier; nevertheless results from acute toxicity, genotoxicity, immunotoxicity, neurotoxicity and reproductive toxicity investigations in different animal models do not provide a clear overview on ION safety yet, and epidemiological studies are almost inexistent. Much work has still to be done to fully understand how these nanomaterials interact with cellular systems and what, if any, potential adverse health consequences can derive from ION exposure.

The importance of organizational commitment to knowledge management

In recent years, Knowledge Management (KM) has assumed great importance in the literature on business and management. However, we still have so little understanding of the human issues in KM processes. Thus, this research aims to contribute to analysing the importance of Organizational Commitment (OC) to KM. First, we used the Cardoso (2003) Knowledge Management Questionnaire (QGC) that embraces all organizational activities around knowledge processes and distinguishes four dimensions of KM. Secondly we applied the Quijano, Masip, Navarro and Aubert (1997) questionnaire (ASH-ICI) that distinguishes two types of commitment (personal and instrumental) into four dimensions. These two questionnaires were applied with 300 employees in the Portuguese industrial ceramics sector. Through multiple regression analysis we found that levels of organizational commitment are statistically important to KM dimensions. Furthermore, our analysis indicates that personal commitment is more important than need commitment. These results are discussed and Organizational Behaviour specialists and Work and Organizational psychologists are challenged to assume more responsibility and an active role in KM studies and practices and to explore human issues in this field.

Building genre knowledge: a case study of efl student perceptions of the argument essay

Dissertação de mest. em Didáctica das Línguas e Culturas Modernas Especialização Inglês, Faculdade de Ciências Humanas e Sociais, Univ. do Algarve, 2003

Knowledge management education:five Ws and one H

This panel will discuss key aspects of knowledge management (KM) education in response to challenges posed by the necessity to improve KM as a discipline and an established professional field. Through panelists' thought-provoking presentations and interactions with the audience, the discussion will address KM education from the starting why, what, who, where and when perspectives to the end result and understanding of how to approach KM education in the future.

The Weight of Language and Action: Epistemic Stance, Methodological Action,and Theoretical Perspective in Knowledge Organization

Knowledge organization (KO) research is a field of scholarship concerned with the design, study and critique of the processes of organizing and representing documents that societies see as worthy of preserving (Tennis, 2008). In this context we are concerned with the relationship between language and action.On the one hand, we are concerned with what language can and does do for our knowledge organization systems (KOS). For example, how do the words NEGRO or INDIAN work in historical and contemporary indexing languages? In relation to this, we are also concerned with how we know about knowledge organization (KO) and its languages. On the other hand, we are concerned with how to act given this knowledge. That is, how do we carry out research and how do we design, implement, and evaluate KO systems?It is important to consider these questions in the context of our work because we are delegated by society to disseminate cultural memory. We are endowed with a perspective, prepared by an education, and granted positions whereby society asks us to ensure that documentary material is accessible to future generations. There is a social value in our work, and as such there is a social imperative to our work. We must act with good conscience, and use language judiciously, for the memory of the world is a heavy burden.In this paper, I explore these two weights of language and action that bear down on KO researchers. I first summarize what extant literature says about the knowledge claims we make with regard to KO practices and systems. To make it clear what it is that I think we know, I create a schematic that will link claims (language) to actions in advising, implementing, or evaluating information practices and systems.I will then contrast this with what we do not know, that is, what the unanswered questions might be (Gnoli, 2008 ; Dahlberg, 2011), and I will discuss them in relation to the two weights in our field of KO.Further, I will try to provide a systematic overview of possible ways to address these open questions in KO research. I will draw on the concept of elenchus - the forms of epistemology, theory, and methodology in KO (Tennis, 2008), and framework analysis which are structures, work practice, and discourses of KO systems (Tennis, 2006). In so doing, I will argue for a Neopragmatic stance on the weight of language and action in KO (Rorty, 1982 ; 2000). I will close by addressing the lacuna left in Neopragmatic thought – the ethical imperative to use language and action in a particular good and moral way. That is, I will address the ethical imperative of KO given its weights, epistemologies, theories, and methods. To do this, I will review a sample of relevant work on deontology in both western and eastern philosophical schools (e.g., Harvey, 1995).The perspective I want to communicate in this section is that the good in carrying out KO research may begin with epistemic stances (cf., language), but ultimately stands on ethical actions. I will present an analysis describing the micro and the macro ethical concerns in relation to KO research and its advice on practice. I hope this demonstrates that the direction of epistemology, theory, and methodology in KO, while burdened with the dual weights of language and action, is clear when provided an ethical sounding board. We know how to proceed when we understand how our work can benefit the world.KO is an important, if not always understood, division of labor in a society that values its documentary heritage and memory institutions. Being able to do good requires us to understand how to balance the weights of language and action. We must understand where we stand and be able to chart a path forward, one that does not cause harm, but adds value to the world and those that want to access recorded knowledge.

Organizational Cooperation and Knowledge Management in Research and Development Organizations

Organizational Cooperation (OC) is a current concept that responds to the growing interdependence among individuals and teams. Likewise, Knowledge Management (KM) accompanies specialization in all sectors of human activity. Most KM processes are cooperation-intensive, and the way both constructs relate to each other is relevant in understanding organizations and promoting performance. The present paper focuses on that relationship. The Organizational Cooperation Questionnaire (ORCOQ) and the Short form of the Knowledge Management Questionnaire (KMQ-SF) were applied to 639 members of research and development (R&D) organizations (Universities and Research Institutes). Descriptive, correlational, linear multiple regression and multivariate multiple regression analyses were performed. Results showed significant positive relationships between the ORCOQ and all the KMQ-SF dimensions. The prediction of KMQ-SF showed a large effect size (R2 = 62%). These findings will impact on how KM and OC are seen, and will be a step forward in the development of this field.

“Biomarkers in a One Health perspective: current knowledge of their ability to merge human and animal health

Biomarkers are nowadays essential tools to be one step ahead for fighting disease, enabling an enhanced focus on disease prevention and on the probability of its occurrence. Research in a multidisciplinary approach has been an important step towards the repeated discovery of new biomarkers. Biomarkers are defined as biochemical measurable indicators of the presence of disease or as indicators for monitoring disease progression. Currently, biomarkers have been used in several domains such as oncology, neurology, cardiovascular, inflammatory and respiratory disease, and several endocrinopathies. Bridging biomarkers in a One Health perspective has been proven useful in almost all of these domains. In oncology, humans and animals are found to be subject to the same environmental and genetic predisposing factors: examples include the existence of mutations in BR-CA1 gene predisposing to breast cancer, both in human and dogs, with increased prevalence in certain dog breeds and human ethnic groups. Also, breast feeding frequency and duration has been related to a decreased risk of breast cancer in women and bitches. When it comes to infectious diseases, this parallelism is prone to be even more important, for as much as 75% of all emerging diseases are believed to be zoonotic. Examples of successful use of biomarkers have been found in several zoonotic diseases such as Ebola, dengue, leptospirosis or West Nile virus infections. Acute Phase Proteins (APPs) have been used for quite some time as biomarkers of inflammatory conditions. These have been used in human health but also in the veterinary field such as in mastitis evaluation and PRRS (porcine respiratory and reproductive syndrome) diagnosis. Advantages rely on the fact that these biomarkers can be much easier to assess than other conventional disease diagnostic approaches (example: measured in easy to collect saliva samples). Another domain in which biomarkers have been essential is food safety: the possibility to measure exposure to chemical contaminants or other biohazards present in the food chain, which are sometimes analytical challenges due to their low bioavailability in body fluids, is nowadays a major breakthrough. Finally, biomarkers are considered the key to provide more personalized therapies, with more efficient outcomes and fewer side effects. This approach is expected to be the correct path to follow also in veterinary medicine, in the near future.