Search

Text data constitutes a significant part of all data generated on the Internet, including the social network users’ comments and posts. Each website offers its users different functionalities. LinkedIn mainly focuses on the labor market as well as professional and business contacts, and Facebook offers the possibility of creating groups as well as photo and message sharing with friends, while Twitter allows short text message posting and tracking. One type of information researchers would like to obtain about the users of these portals is their age. Such information is crucial from the perspective of marketing, social and economic research. Each of the social networks, however, has different rules regarding the privacy policy and the publishing of information about the date of birth. This poses a problem for the researchers who would like to obtain such information. The aim of the research presented is to attempt characterization of the words typically used in the messages published by Twitter users. This social networking site was chosen due to the possibility of downloading data without additional user consent. Text mining methods and techniques were used to carry out the research, which was mainly focused on the analysis of individual words and collocations occurring in the users’ tweets.

The increasing amount of digital data and the declining cost of data storage have led to the fact that companies began collecting any the data possible, regardless of its adequacy and usability. This results in increasingly diverse data, in terms of its structure, quality, availability and the source of origin. Dark data is one type of data that increases significantly as the volume of data expands. Scientific literature does not precisely define the term “dark data”, while its interpretation among scientists is ambiguous. The aim of this article entails an attempt to define the dark data occurring in an enterprise, by identification of its essential features. The article presents an overview of the definitions of the term dark data, a proposal of its interpretation, and a classification of data in a company with regard to: usability, availability and quality. The analysis of the concept of dark data was carried out via a review of international journals and articles published on the Internet by Data Science practitioners. As part of the research, four universal features of dark datasets have been indicated (unavailability, unawareness, uselessness, and costliness). Based on data availability and its quality, four groups of enterprise data have also been distinguished. The data classification developed in this way allowed systematization of the term “dark data”.