Graduation Thesis Sample

Big Data and Semantic AnalysisIn this thesis several relatively new concepts will be mentioned that havealready begun changing the scope of today's business. These are the Big Dataand semantic analysis concepts. In these thesis, the potential of this topic wasexplored, and this work expanded these two terms and linked them toconcrete applications in business organizations. Although authors andresearchers still cannot agree on what is the specific definition of the termBig Data, often in the literature, in an effort to describe the complexity of thisterm, the so-called V's approach is mentioned. Most authors, like those whowill be quoted in this paper, use 4 V: Volume, Variety, Velocity and Veracity.Big Data solutions are ideal for analysis of not only structured data, whichbusiness organizations are used to analyze, but also unstructured and semi-structured data that often come from different sources. In this paper, specialattention will be paid to unstructured data. Specifically, textual data fromsocial networks and popular websites will be researched. Large data isconsidered to be ideal when it is necessary to analyze all data that isconsidered relevant for better understanding of clients. The other term referred to is semantic analysis. The goal of semantic analysisis to understand the meaning of a particular linguistic input. Therefore, thedata is collected, the text is converted into a number, and the obtainedresults are used in further business analysis, which leads to an increase inthe value of existing analyzes and outputs, since these data were unavailableto us (at least small and medium-sized enterprises). 

Semantics deals with the analysis of meaning and stands at the center of alinguistic quest to understand the nature of language and linguistic abilities.Sentiment analysis or analysis of thinking is a field of science that analyzeshuman thoughts, feelings, praise, attitudes and emotions towards differentproducts, services, organizations, people, problems, events and theirattributes. Therefore, in this paper, the semantic analysis will analyze theopinions of people published on social networks and websites. Bothconcepts (Big Data and Semantic Analysis) have been known and haveexisted for quite some time, but in recent years, with the development of BigData technology, the price of this kind of analysis has been reduced, and thepotential of unstructured data has been exploited. Particular attention shouldbe paid to the economic viability of this type of unstructured data (textualdata) in modern business. Most authors deal with the technological problemand the technological feasibility of this type of analysis, while the economicaspect is often unfairly ignored.Research subjectToday, companies are increasingly paying attention to the data-driven wayof thinking and doing business. That is, their decisions are driven by data.Data needs considerably increase; companies require more and betterquality and more diversified data, with the aim of extending their analysisand gaining a wider view of their customers. The question arises: is itpossible to get quality data that can contribute to decision-making inmodern business? The contribution will be analyzed through a prism oftechnological and economic approach. The Big Data technologies thatsupport this kind of analysis will be analyzed, as well as the models forsemantic analysis and selection of optimal and usability for businessdecision making. 

In order to analyze the state of this type of analysis, a concrete projectfinanced by the European Union, which meets all the above criteria, will beused. The entire path required for semantic analysis will be processed. Fromcollecting data, saving it, ETL, modeling, creating outputs, and utilizingoutputs to make decisions. Frequently, the problem is that during theimplementation, the moment of exploitation of the decision-making outputis reached, with often delaying or deterioration of the project itself. Sourcesof this problem will also be identified in this paper.Research hypothesesAn important part of the paper is dedicated to setting up appropriateresearch hypotheses. Hypothesis (Greek hypothesis, assumption) is theacceptance of the assumption on which a conclusion is based, which servesfor advancement of research and explanation, without being proven byother principles and not confirmed by (verified) experience. Therefore, thegoal is to prove, or not to reject, hypotheses.The following research hypotheses should be based on the applicativeresearch, verify truthfulness. The hypotheses are:Ho: Semantic analysis of unstructured data supported by Big Datatechnology is usable for business decision makingH1: Semantic analysis of unstructured data supported by Big Datatechnology is not usable for business decision-makingTo determine whether the semantic analysis of unstructured data issupported by Big Data is usable for business decision-making, it is necessaryto determine whether the model obtained by semantic analysis has sufficientquality output, which can be used for business decision-making. 

Therefore, in this paper we will examine these sub-hypotheses:Ho: Data obtained by semantic analysis of unstructured data supported byBig Data technology is of high quality.H1: Data obtained by semantic analysis of unstructured data supported byBig Data technology is not of high quality. The quality of data obtained by semantic analysis of unstructured datasupported by Big Data technology will be determined by:- The time it takes from the start of the analysis to the creation of the outputbased on analyzes,- The amount of resources needed for this type of analysis- Using the accuracy of the model.Research aimThe aim of the research is to confirm the hypotheses. By interpreting thehypotheses, the goals are reduced to the conclusion that the semanticanalysis of unstructured data is supported by Big Data technology usable forbusiness decision making. The achievement of objectives will be achieved byapplying the methodological framework, which is explained in more detailin the next chapter. The backbone of the research will be the implementationof applied research on a concrete example of the project and evaluating theresults obtained. The results obtained should be used in order to betterunderstand the maturity of Big Data technology and semantic analysis oftextual data for the delivery of quality data for business decision-makingpurposes on the example of this company's project.

ReferencesAggarwal Charu C. & Zhai, C. X. (2012). Mining Text Data, USA: Springer.Richert, W. & Coelho, L. P. (2013). Building Machine Learning Systems withPython, UK: Packt Publishing.Harris, H. et al. (2013). Analyzing the Analyzers, an Introspective Survey ofData Scientists and Their Work, USA: O'Reilly.