Search

Presents the introductory welcome message from the conference proceedings. May include the conference officers' congratulations to all involved with the conference event and publication of the proceedings record.

Presents the introductory welcome message from the conference proceedings. May include the conference officers' congratulations to all involved with the conference event and publication of the proceedings record.

Presents the introductory welcome message from the conference proceedings. May include the conference officers' congratulations to all involved with the conference event and publication of the proceedings record.

Big Data has become a new source of opportunity among applications in Artificial Intelligence. Many design considerations exist in this relatively new field where parallel processing frameworks can be employed in a more economical fashion. Unlike traditional data sources, Big Data applications present their own unique challenges in order to appropriately harness the utility of open source frameworks including Apache Spark and design patterns predicated on the Directed Acyclic Graph. By embracing this new paradigm, parallel processing can be effectively leveraged to support development at a level of scale and performance that was not possible earlier.

Semantic Web standards were developed in order to improve accessibility to the web. While developers frequently leverage software APIs to support application development, considerable challenges still exist. The Semantic Web has promised to resolve these challenges, but to date has only provided limited functionality. Higher potential has been demonstrated in the area known as Linked Data which has leveraged Semantic Web technologies to support the purpose of building applications by linking data sets within a common model. This paper examines a number of potential applications for mobility which leverage Linked Data thereby providing direct access to digital libraries and information resources.

Big Data has become ubiquitous across all areas of research allowing for new applications that were not possible earlier. Unlike software development relying on traditional data sources, Big Data applications present their own unique challenges to appropriately harness the utility of the Apache Hadoop architecture. In this paper, we introduce fundamental concepts of Hadoop and explore its usage as well as future direction. We also present our strategy for exploring the Hadoop architecture including addressing issues of scalability, customization of code and utilization of programming techniques.

The ever-increasing amount of information flowing through Social Media presents numerous opportunities for the generation of Business Intelligence. Challenges exist in the leveraging of these data sources due to their heterogeneity and unstructured content. This paper presents the application of Semantic Computing to Social Media for industrial application, focusing on topic identification and behavior prediction. The methodologies described can benefit many areas of an organization including support of marketing, customer service, engineering and public relations. Results demonstrate that business operations can be substantially enhanced through application of Semantic Computing to Social Media.

With rapidly expanding data collections becoming increasingly available, the application of Semantic Computing has become imperative to leverage this resource for industrial applications. This paper presents a survey of Semantic Computing in the area of Business Intelligence. We examine semantic analytical techniques and tools as applied for prediction analysis and decision support. We also define the role of Semantic Computing as applied in the context of Data Mining, Text Mining and Big Data Analytics. Additionally, we describe how business data is queried with Structured Natural Language as well as the use of On-Line Analytic Processing.

One of the most substantial opportunities in Big Data is to integrate disparate data sources across the enterprise. To realize this goal it can be valuable to leverage environments developed for high speed parallel processing as well as toolkits supporting the development and maintenance of semantic information. In support of this approach a proposed framework and methodology is presented to utilize an ontology-based data integration strategy. Our approach supports a rule-based translation to generate new ontology versions within a fast prototyping environment leveraging the Jena API within the context of the Apache Spark environment.

In order to realize the promise of Big Data, applications will have to consider the integration of many disparate data sources. Due to data heterogeneity, this task presents a number of challenges that may not be completely resolved with existing Extract-Transform-Load (ETL)-based frameworks. In this paper, we explore the potential of Semantic Web Technologies as a means of integration and development of Big Data applications. In demonstration, a usage case study is presented examining supplier chain operations. Additionally, we review the overall challenges of data integration in this area.

A challenge in the research of Social Networks is the large scale analysis of graphs. One of the most valuable metrics in the evaluation of graphs is betweenness-centrality. In this paper, we define an approximation of betweenness-centrality for the purpose of building a predictive model of Social Networks. The methodology presented describes a bounded distance approximation of betweenness-centrality designed for implementation within a parallel architecture. Through our proposed design pattern, we are able to leverage Big Data technologies to determine metrics in the context of ever expanding internet-based data resources.

Due to its predictive nature, Social Media has proved to be an important resource in support of the identification of trends. In Customer Relationship Management there is a need beyond trend identification which includes understanding the topics propagated through Social Networks. In this paper, we explore topic modeling by considering the techniques of Latent Dirichlet Allocation which is a generative probabilistic model for a collection of discrete data. We evaluate this technique from the perspective of classification as well as identification of noteworthy topics as it is applied to a filtered collection of Twitter messages. Experiments show that these methods are effective for the identification of sub-topics as well as to support classification within large-scale corpora.

Big Data has become a new source of opportunity in terms of Semantic Computing. Unlike traditional data sources, Big Data applications present their own unique challenges in order to appropriately harness the utility of open source frameworks including Apache Hadoop and design patterns such as Map - Reduce. Through these advances, Big Data has become ubiquitous across all areas of research allowing for new applications that were not possible earlier. The Second IEEE International Workshop on Semantic Computing with Big Data (ICSC-SCBD'15) provides a forum for researchers to present their latest research progress, exchange ideas and thoughts, and explore new research directions. It continues seeking high quality research contributions and encouraging significant work that addresses the major challenges in the semantics as applied in the context of Big Data.

Twitter-based messages have presented challenges in the identification of features as applied to classification. This paper explores filtering techniques for improved trend detection and information extraction. Starting with a pre-filtered source (Twitter), we will examine the application of both information theory and Natural Language Processing (NLP) based techniques as a means of preprocessing for classification. Results demonstrate that both means allow for improved results in classification among highly idiosyncratic data (Twitter).

The MapReduce paradigm has become ubiquitous within Big Data Analytics. Within this field, Social Networks exist as an important area of applications as it relies on the large scale analysis of graphs. To enable the scalability of Social Networks, we consider the application of MapReduce design patterns for the determination of graph-based metrics. Specifically, we detail the application of a MapReduce-based solution for the metric of betweenness-centrality. The prevailing concept is separation of the graph topology from the actual graph analysis. Here, we consider the chaining of MapReduce jobs for the estimation of shortest paths in a graph as well as post processing statistics. Through our design pattern, we are able to leverage Big Data Technologies to determine metrics in the context of ever expanding internet-based data resources.

Applications that utilize publically available content from the web have been successful in tracking major events across a number of areas. We have developed a method of filtering to characterize trends of consumer behavior in relationship to specific products using the Twitter messaging system. Our process considers semantics at three successive levels to determine a demand signal. This begins with the establishment of ground truth keywords followed by word-level and category-level empirical keywords. Next, semantic categories of humor, emotion and negation are considered. Following, a classifier is applied for additional filtering to further support the characterization of consumer behavior. We apply this procedure to the goal of modeling vehicle purchase behavior with data acquired from Twitter. Results present strong correlation to sales data, allowing for contributions to forecasting efforts as well as Customer Relationship Management (CRM).

This paper considers the extraction and analysis of Social Networks for the identification of trends. Our methodology focuses on the utilization of semantics for determination of relevant networks within unstructured data. The Social Networks are examined from the perspective of structure and considered as a time series. Our metrics focus on the identification of influence and power among key players. This method is applied against a collection of Twitter messages and compared to historical market share trends of technologically-related topics. Through this work we demonstrate that structural qualities reflecting community dynamics can provide insight to the prediction of long-term trends. The goal of this work is to lend insight to the characterization of consumer behavior, particularly in the area of technology forecasting.

Social networks today represent a substantial amount of shared knowledge and information. To leverage the interdependence of this data, we consider two forms of relational learning to facilitate semantic understanding. First, relational modeling is applied to local networks to reinforce knowledge in each entity. Then, a social dimension approach is applied to generate new (high level) features. These feature sets are then trained towards the identification of learned purchase behaviors (belief system / values) thus supporting a means of prediction. We consider this generation of higher level classifications (termed as social dimensions) to enable increased accuracy in behavior prediction in order to support more focused customer relationships.

Social media has demonstrated itself to be a proven source of information towards the marketing of products. This unique source of data provides a rapid means of customer feedback that is used to support a number of business areas. Towards this purpose, we describe a methodology for the identification of topics associated with customer sentiment. This process first employs a Fisher Classification based approach towards sentiment analysis. By considering specific mutual information and word frequency distribution, topics are then identified within sentiment categories. The goal is to provide overall trends in sentiment along with associated subject matter (ie. why) as it supports a company's business. We demonstrate this methodology against data collected among a particular product line as obtained from Twitter advanced search.

Increased sources and quantity of unstructured information has created a further need for categorization and interpretation of their content. This paper describes the design of an interchangeable framework to support learning from an unstructured data source. Our approach supports integration of two or more learning mechanisms with a traditional indexing method. The goal is to identify a higher semantic content and more meaningful keyword combinations, considering both supervised and unsupervised techniques. Within a specific implementation both Bayesian learning as well as Clustering are integrated to support a boost parameter towards classification of unstructured text. We find that an implementation of this framework applied towards a set of Reuters news feeds supports a vastly improved recognition rate. Our effort is directed towards making associations between structured and unstructured information.

Machine Learning algorithms are difficult to directly apply among data sets of high dimensionality. This paper examines application of hybrid algorithms to segment data models to enable a higher level of accuracy. Our process begins with the reduction of our input parameter sets through the derivation of dominant characteristics. Using these characteristics, ranges are determined in which to segment our model. Each set is then used to train a predictive model using Machine Learning techniques. One major attribute of our application framework is to support an interchangeable set of algorithms for each stage. This process is demonstrated by estimating stated incomes from an automotive financing application for purpose of predictive modeling. We conclude that by applying our segmented hybrid framework we can achieve substantial improvements in accuracy over pure Machine Learning applications.

This paper presents a rule based approach to ontology refactoring. Our method supports large scale instance relationships for support of translation to an improved, functionally equivalent design. By generating new ontology versions on-the-fly we can verify potential updates to our requirements. An example of this technique is presented utilizing a subset of the OWL-DL specification through the implementation of the Jena API. Advantages of this approach include rapid prototyping, versioning support, querying and tool development to support the automated engineering of instance data.

This paper presents framework to support design of meta-rule constructs. A prototype is described towards the application of credit analysis. The focus of this system is to define a higher level of inference that will guide pre-established object-level rule constructs. This architecture is supported by the incorporation of machine-learning (ML) techniques to support the acquisition of business rule-based knowledge. This knowledge is applied to specific parameters that ultimately guide an object level decision making process. By developing this process of automated knowledge acquisition, we are interested in up front actions including validation and support of intended responses. We also intend to further classify and categorize this acquired knowledge to support future policy modifications.

This paper presents an approach to ontology development through the application of declarative logic programming. Our method employs rules for the purpose of prototyping new ontology versions by decoupling the process of concept definition from the application of descriptive logics (DL) and advanced class representations. By generating new ontology versions on-the-fly we can test updates to the ontology design. This employment of rules expands on current efforts of translation and merging of ontologies. By employing this technique, we can support a pragmatic approach to the management and integration of instance data thus realizing a rapid-prototyping approach to the testing of potential updates to ontologies. Examples of this technique are presented utilizing a subset of the OWL-DL specification through the implementation of the Jena API. Advantages include the rapid testing of updated ontology representations (including the efficient remapping of instance data) and an efficient means of Ontology querying. Eventual benefits include Ontology versioning support and tool development to support the automatic engineering of instance data.

Cultural algorithms are computational self-adaptive models consisting of a population and a belief space. Two cultural algorithms are applied, with one supporting white box testing and the second black box testing. The two populations communicate with each other by means of a shared belief space. This is applied to the calibration of a multi-agent system by allowing for evolution of near optimal parameters. The cultural approach is employed to abstract coefficients of pricing strategies that are applied to a complex model of durable goods. This model simulates consumer behaviors as applied in the context of economic recession.

Cultural algorithms are computational, self-adaptive models consisting of a population and a belief space. In this framework, a white and black box testing strategy is embedded in order to test large-scale GP programs. The model consists of two populations, one supporting white box testing of a genetic programming system and the other supporting black box testing. The two populations communicate with each other by means of a shared belief space. The white box component is examined in detail. This component is first demonstrated by evolving a non-linear function. The cultural white box approach is then employed to abstract coefficients of pricing strategies that are applied to a complex model of durable goods. This model simulates consumer behaviors as applied in the context of economic cycles.

Evolutionary computation has been successfully applied in a variety of problem domains and applications. In this paper we discuss the use of a specific form of evolutionary computation known as cultural algorithms that has been applied successfully in various real-world applications to solve problems of a very dynamic and complex nature. Cultural algorithms introduce a learning component into an evolutionary framework that influences the search strategy and is in turn modified by the best-performing members of the population during the entire process. Cultural algorithms have been used in various applications, including fraud analysis for automotive accident claims, the re-engineering of a dynamic automobile manufacturing knowledge base, the modeling of pricing strategies for automobiles in a multi-agent environment and for data mining.

Cultural algorithms are self-adaptive models that support the collective evolution process through the employment of a population and a belief space. The cultural approach is applied to derive a generalized set of beliefs from successive populations of parameter configurations from an agent-based simulation of transactions within a durable goods market. The maintenance of this information allows for the guided evolution of the agent-based system over successive simulations. In order to more effectively evaluate parameter configurations, software engineering techniques of white and black box testing are applied. In this paper, a methodology for the use of cultural algorithms to optimize strategies in agent-based models is presented. This approach is demonstrated in an application used to model pricing strategies in the context of an agent-based model under a simulated real-world market scenario and a heterogeneous population.

Companies' risk due to foreign exchange (FX) rates can be reduced using market-based instruments. We have used the value-at-risk methodology to evaluate FX risk, and have developed a genetic algorithm-based optimization framework for evaluating different strategies. The best strategy is to hedge an identical percentage of each FX exposure.

Software testing is extremely difficult in the context of large scale engineering applications. We suggest that the application of the white and black box testing methods within a cultural algorithm environment will present a successful approach to fault detection. In order to utilize both a functional approach and a structural approach, two cultural algorithms will be applied within this tool. The first cultural algorithm will utilize the black box testing by learning equivalence classes of faulty input/output pairs. These equivalence classes are then passed over to the second cultural algorithm that will apply program slicing techniques to determine program slices from the data. The goal will be to pinpoint specific faults within the program design. Through the searching of the program code, this approach can be considered as behavioral mining of a program.

In this paper, we use Cultural Algorithms as a framework in which to embed a white and black box testing strategy for designing and testing large-scale GP programs. The model consists of two populations, one supports white box testing of a genetic programming system and the other supports black box testing. The two populations communicate by sending information to a shared belief space. This allows a potential synergy between the two activities. Next, we exploit this synergy in order to evolve an OEM pricing strategy in a complex agent-based market environment. The new pricing strategy generated over $2 million dollars in revenue during the assessment period and outperformed the previous optimal strategy.

Programmers often employ knowledge-based heuristic approaches in the application of solving programming problems. Program slicing is one tool used to acquire such knowledge within the area of Software Engineering to support the debugging, testing, maintenance and understanding of programs. Program slicing is the determination of the set of all the statements in a program that directly or indirectly affects the value of a variable occurrence. Genetic Programming is the process of using evolutionary techniques to identify information that can be used to identify the location of problems in program code. We believe that within a Cultural Algorithm framework, a testing analysis agent can be implemented utilizing slicing techniques in order to produce more accurate program metrics.