Conceptual Structures for STEM Research and Education

Linked and open data is increasing being used by governments, business and administration. Awareness of the affordances and potential utility of open data is being raised by the emergence of a host of web-based and mobile applications.

Across the educational and research communities applications applying the principles linked data principles have emerged.

Systems developed and used by researchers and academics are most likely to be predominantly in the hands of the early adopters and current developments found in higher education tend to be atomized, yet there is potentially considerable advantage in associating and integrating applications for organisational, educational and administrative.

This paper presents an argument for how we can move from early adopters to early majority, and at the same time presents a roadmap which will outline some of the significant challenges which remain to be addressed.

The world is a continuum, but words are discrete. Sensory organs map the continuous world to continuous mental models of sights, sounds, and motions. Muscles and bones move in a continuous range of positions, postures, forces, and speeds. Internal feelings of hunger, thirst, pains, pleasures, fears, and desires have a continuous range of variation. But discrete words and patterns of words cannot faithfully represent the continuum of perceptions, actions, and feelings. Peirce’s semiotics and Wittgenstein’s language games provide a framework for relating language to the world and to perceptions and actions in the world. Peirce analyzed signs and transformations of signs in networks of discrete symbols and in patterns of continuous images. Wittgenstein showed how language is integrated with every aspect of human activity. To implement their insights, the discrete networks of symbols must be mapped to continuous mathematics. This article is a summary of the methods and applications for mapping natural languages to conceptual graphs and continuous transformations. Those methods have been used to analyze and classify plot twists in narratives and the structure of expository texts.

This paper focuses on the challenges of introducing classical syllogisms in university courses in elementary logic and human reasoning. Using a program written in Prolog+CG, some empirical studies have been carried out involving three groups of students in Denmark; one group of philosophy students and two groups of students of informatics. The skills of the students in syllogistic reasoning before and after the logic courses have been studied and are discussed. The empirical observations made with the program make it possible to identify syllogisms which are found difficult by the students, and to identify others which the students find easier to handle. It is discussed why certain syllogisms are more difficult than others to assess correctly with respect to validity. The results are compared with findings from earlier studies in the literature. As in other studies, it is shown that the test persons have a tendency correctly to assess valid syllogisms as such more often than correctly assessing invalid syllogisms as such. It is also investigated to what extent the students have improved their skills in practical reasoning by attending the logic courses. Finally, some open questions regarding syllogistic reasoning are discussed.

In our previous work we have introduced a novel social media that performs collaborative filtering on situations. This enhances user situation awareness with a collaborative effort to learn about importance of situations. In this paper we focus on defining a conceptual graph-based model used to represent situations in our system, so that it would (1) be consistent with existing formal definitions of situation, and (2) enable logical manipulations on situations, namely their detection and semantic generalization, which we employ in the system. In particular, we show how the latter can be accomplished thanks to situation lattices, which we adapt for the model.

The analysis of big data, although potentially a very rewarding task, can present difficulties due to the complexity inherent to such datasets. We suggest that conceptual graphs provide a mechanism for representing knowledge about a domain that can also be used as a useful scaffold for big data analysis. Conceptual graphs may be used as a means to collaboratively build up a robust model forming the skeleton of a data analysis project. This paper describes a case study in which conceptual graphs were used to underpin an exploration of a corpus of tweets relating to the Transportation Security Administration (TSA). Through this process we will demonstrate the emerging model built up of the data landscape involved and of the business structures that underlie the technical frameworks relied upon by microblogging software.

The SWIP system aims at hiding the complexity of expressing a query in a graph query language such as SPARQL. We propose a mechanism by which a query expressed in natural language is translated into a SPARQL query. Our system analyses the sentence in order to exhibit concepts, instances and relations. Then it generates a query in an internal format called the pivot language. Finally, it selects pre-written query patterns and instantiates them with regard to the keywords of the initial query. These queries are presented by means of explicative natural language sentences among which the user can select the query he/she is actually interested in. We are currently focusing on new kinds of queries which are handled by the new version of our system, which is now based on the 1.1 version of SPARQL.

With the growth of interest in semantics around the world, we believe that conceptual graphs have an important role to play. However, from the best of our knowledge, there is a lack of conceptual graphs storage and retrieval engine capable of scaling up. In this paper, we propose to utilize the power of the RDF stores, and present a complete system and methods to implement an efficient conceptual graphs storage and retrieval engine in an RDF store. We translate conceptual graphs knowledge bases into RDF knowledge bases, create an external index of the conceptual graphs and use the index to efficiently retrieve a set of candidate conceptual graphs in response to an expanded user query. We also discuss several heuristics which aim to speed up the data retrieval process, and present preliminary experimental results using the different heuristics.

This paper discusses the notion of medical archetype and the manner how the archetype elements are documented in hospital patient records. This is done by interpreting the archetypes as information extraction templates in automatic text analysis of clinical narratives. The extensive extraction experiments performed over thousands of anonymous discharge letters show the actual instantiation of the required and expected items in the narrative clinical documentation; in fact much tacit medical knowledge is implicitly presented in the real clinical texts. This fact suggests that the archetype approach to defaults and inheritance might need certain development.

This paper discusses the use of conceptual structures in the design of computer-based assessment (CBA) tools for e-assessment of programming exercises. In STEM (science, technology, engineering and maths) subjects, universities often observe high dropout and failure rates among the first year students. There are a number of research initiatives that investigate the use of interactive teaching methods and e-learning technologies for improving STEM education. This paper presents a conceptual model of programming exercises and discusses more generally how conceptual structures can be employed for the implementation of CBA tools.

In the domain of ontology design as well as in Conceptual Modeling, representing universals is a challenging problem. Most approaches which have addressed this problem rely either on Description Logics (DLs) or on First Order Logic (FOL), but many difficulties remain especially about expressiveness. In mathematical logic and program checking, type theories have proved to be appealing but so far, they have not been applied in the formalization of ontologies. To bridge this gap, we present here the main capabilities of a theory for representing ontological structures in a dependently-typed framework which relies both on a constructive logic and on a functional type system. The usability of the theory is demonstrated with the Coq language which defines in a precise way what ontological primitives such as classes, relations, properties and meta-properties, are in terms of type classes.

We develop a graph representation and learning technique for parse structures for sentences and paragraphs of text. This technique is used to improve relevance answering complex questions where an answer is included in multiple sentences. We introduce Parse Thicket as a sum of syntactic parse trees augmented by a number of arcs for inter-sentence word-word relations such as coreference and taxonomic. These arcs are also derived from other sources, including Rhetoric Structure theory, and respective indexing rules are introduced, which identify inter-sentence relations and joins phrases connected by these relations in the search index. Generalization of syntactic parse trees (as a similarity measure between sentences) is defined as a set of maximum common sub-trees for two parse trees. Generalization of a pair of parse thickets to measure relevance of a question and an answer, distributed in multiple sentences, is defined as a set of maximal common sub-parse thickets. The proposed approach is evaluated in the product search domain of eBay.com, where user query includes product names, features and expressions for user needs, and the query keywords occur in different sentences of text. We demonstrate that search relevance is improved by single sentence-level generalization, and further increased by parse thicket generalization. The proposed approach is evaluated in the product search domain of eBay.com, where user query includes product names, features and expressions for user needs, and the query keywords occur in different sentences of text.

This paper considers a data analysis system for collaborative platforms which was developed by the joint research team of the National Research University Higher School of Economics and the Witology company. Our focus is on describing the methodology and results of the first experiments. The developed system is based on several modern models and methods for analysing of object-attribute and unstructured data (texts) such as Formal Concept Analysis, multimodal clustering, association rule mining, and keyword and collocation extraction from texts.

Belief Revision was conceived to model how humans do think, and has found application in machine learning. This paper argues that Peirce’s theory of inquiry conceives how we must think, if we want to keep improving our knowledge. Distinguishing between these two views, psychological (empirical) and pragmatic (normative), is crucial to our improvement of human learning methodology, especially as we develop interactive engagement methods for learning STEM concepts. Examining efforts to model Belief Revision in AI can reveal the limitations of this conceptualization for human learning, due to its misconception of Peirce’s pragmatic theory of inquiry.

This paper describes the first-order logical environment FOLE. Institutions in general (Goguen and Burstall [4]), and logical environments in particular, give equivalent heterogeneous and homogeneous representations for logical systems. As such, they offer a rigorous and principled approach to distributed interoperable information systems via system consequence (Kent [6]). Since FOLE is a particular logical environment, this provides a rigorous and principled approach to distributed interoperable first-order information systems. The FOLE represents the formalism and semantics of first-order logic in a classification form. By using an interpretation form, a companion approach (Kent [7]) defines the formalism and semantics of first-order logical/relational database systems. In a strict sense, the two forms have transformational passages (generalized inverses) between one another. The classification form of first-order logic in the FOLE corresponds to ideas discussed in the Information Flow Framework (IFF [12]). The FOLE representation follows a conceptual structures approach, that is completely compatible with formal concept analysis (Ganter and Wille [2]) and information flow (Barwise and Seligman [1]).

Transactional systems are core to much business activity; however leveraging any advantage from the data in these enterprise systems remains a challenging task for businesses. To research and discover the hidden semantics in transactional data, Sheffield Hallam University has incorporated Formal Concept Analysis (FCA) into two of its degree courses. We present a learning, teaching and assessment (LTA) method that integrates with this research. To make it reflect industrial practice and to further the state of the art of the research, this method includes the use of ERPsim. This large scale, real-world business simulation software is based on the Enterprise Resource Planning (ERP) enterprise system by SAP A.G., a global business software vendor. Together with a mix of individual and group work approaches, FCA tools (namely FCA BedRock, In-Close and Concept Explorer) and comparisons with alternative approaches, it is emerging that FCA can fulfil an important role in transactional systems and enhance its role in Business Intelligence (BI).

Knowledge bases, as conceptual graphs, are considered to be brittle as they are highly domain specific. This paper attempts to get some flexibility by predicting the possible nodes, using the other existing graphs. Graph theory principles of maximum common sub-graph and minimum common super-graph for labelled graphs, allow extension of a given conceptual graph. This paper attempts to solve this problem for laws of science. Given a few fundamental equations of two different domains, but similar mathematical structure,equations can be converted to a common set of dummy variables. These transformed equations will be the labels for further set operations. Extending the two graphs using the minimum common super-graph and maximum common super-graph, we then convert these transformed equations back to their original variables. Then, apply constraints to check the feasibility and finalize this extension. Thus we have inferred some part of the knowledge base from other domains.

We propose a conceptual graph-based framework for abstractive text summarization. While syntactic or partial semantic representations of texts have been used in literature, complete semantic representations have not been explored for this purpose. We use a complete semantic representation, namely, conceptual graph structures, composed of concepts and conceptual relations. To summarize a conceptual graph, we remove the nodes that represent less important content, and apply certain operations on the resulting smaller conceptual graphs. We measure the importance of nodes on weighted conceptual graphs by the HITS algorithm, augmented with some heuristics based on VerbNet semantic patterns. Our experimental results are promising.

This paper aims at pointing out a range of differences between logical form as used in logic and logical form (LF) as used in the minimalist architecture of language. The differences will be shown from different angles based on the ways in which they differ in form and represent some natural language phenomena. The implications as following on from such differences will be then linked to the issue of whether semantic realization in mind/brain is computational. It will be shown that the differences between logical form as used in logic and logical form (LF) as used in the minimalist architecture of language will help us latch on to the realization that there is no determinate way in which semantics can be computational or computationally realized.

The paper describes models for knowledge representation, i.e. extracted at different steps of multidimensional measurements processing procedure, in the context of JDL data fusion model. Models are developed taking into account the requirements of geo information systems environment. As a case study system of conditions lighting which implements the models for oceanographic data processing is described.

Multiple attempts have been made these days to automate the creation of class diagrams by providing structured English statements as input. The resulting diagrams are of close proximity to what the user wants. This paper is one such attempt to transform business designs written in OMG’s (Object Management Group) standard SBVR (Semantics of Business Vocabulary and Rules) framework, into a set of classes in UML (Unified Modeling Language) Class Model using a theoretical approach. SBVR provides a set of specific rules which are processed in order to get class diagram of close proximity. It involves the transformation of “Structured English” into a set of UML Class Model with SBVR as a mediator. Further, the results of the approach are validated using VeTIS tool.

The paper discusses the relation between a novel approach of knowledge engineering, the event bush, and the formalism of directed hypergraph. Despite the seemingly obvious similarity, the relation appears to be far from transparent. However, if formulated accurately, it may give a handy demonstration tool for the event bush approach and open a new avenue of research in directed hypergraph theory.

Conceptual patterns can be described by graphs, entailment by graph homomorphism. The mapping of a pattern to its set of instantiations, represented as a table, constitutes one half of a Galois connection. The join operation is the infimum in a complete lattice of tables, and a most descriptive pattern can be assigned to each table by means of a categorial product construction. This construction constitutes the other half of the Galois connection. In this approach, relational structures assume the role of formal contexts in standard Formal Concept Analysis (FCA). Concepts arise as connected components of powers of these relational structures. The ordered set of these concepts may be conceived as a navigation space.

Median networks have been proposed as an improvement over trees in phylogenetic analysis. This paper argues that concept lattices represent essentially the same information as median networks but with the advantage that there is a larger FCA research community and a variety of available software tools. Therefore evolutionary analysis is an interesting new application domain for FCA.

We present txt2vz (txt2vz.appspot.com), a new tool for automatically generating a visual summary of unstructured text data found in documents or web sites. The main purpose of the tool is to give the user information about the text so that they can quickly get a good idea about the topics covered. Txt2vz is able to identify important concepts from unstructured text data and to reveal relationships between those concepts. We discuss other approaches to generating diagrams from text and highlight the differences between tag clouds, word clouds, tree clouds and graph clouds.