Abstract:

Internet, apart from a huge repository of information of any kind, has become the main means of modern communications and World Wide Web has emerged as a new sort of society since it usually reflects almost all aspects of modern societies in terms of their economic, political and social status and structure. Therein, over wired and wireless connections, through ingenious ideas, i.e., algorithms, that exploit the enclosed computational power, a new kind of culture emerges combing elements from existing traditional civilizations/cultures like for instance history, arts, science and technology, education, language… Motivated by its fundamental and influential nature of the Greek language (see for example the two speeches of Xenophon Zolotas, a Greek economist who served as an interim non-party Prime Minister of Greece in 1989, in English which are considered to be historic because they contained only terms of Greek origin), our paper investigates the influence in written texts hosted in the World Wide Web. Otherwise stated, our work addresses the question: How Greek the web is?
The basic idea of our answer to this question mainly lies in automatically detecting and measuring the frequency (i.e., the percentage) of words of Greek origin in user-selected websites; we focused on websites including English text – but our work can be (easily) extended to websites containing text in other languages. The methodology we used consists in (i) the creation of an exhaustive collection of Greek lemmata, (ii) the design and implementation of a cultural algorithm which starting from an initial set of words of Greek origin collectively learns and detects such words in webpages and (iii) the design and implementation of a user interface which visualizes and presents obtained results.
Cultural algorithms have been suggested for modelling the evolution of cultural systems and involve a belief space, a population, an acceptance function and an influence function. The best individuals of the population can update the belief space via the update function. The knowledge categories of the belief space can affect the population component via the influence function, for example, by altering the genome or the actions of the individuals. In our work, cultural algorithms have been used for building a learning system able to recognize words of Greek origin in WWW texts.
Our application has been implemented using Python, a free, open-source programming environment that supports a clear and expressive syntax offering implementation efficiency and fast execution.
The experimental evaluation of our system using a rich collection of websites from the fields of technology, education and development confirms its learning efficiency and correctness in the qualitative characterization of websites regarding the Greek flavor of their textual content. Such a system could be used in practice as a tool for educational and/or scientific purposes which could include sociological/cultural studies as well as online evaluation of language skills.

Keywords:

Cultural algorithms, python, culture, language, Greek.