Investigation of changes in science textbooks using term frequency-inverse document frequency (TF-IDF) analysis

The aim of this thesis study is based on analyzing the changes in the subject area of physics in science textbooks from 2000 to the present day with automatic text analysis methods. In this study, document analysis method, which is frequently used in quantitative research methods, was used. The sample of the study consists of one textbook from each grade level prepared according to the 2000, 2005, 2013 and 2018 science curricula accepted as textbooks by the Ministry of National Education. In this direction, six (6) textbooks belonging to the 2018 science program were accessed in PDF format via the Ministry of National Education Educational Informatics Network, four (4) books (2000 program 6th, 7th, 8th grade and 2005 program 8th grade) were accessed from the Ankara Ferit Ragıp Tuncor Ministry Archive and Documentation Library serving within the Ministry of National Education, and the remaining twelve (12) books were obtained from bookshops. The books included in the sample were also included in the sample of the study after checking their suitability to the curricula. These books were converted into plain text documents using different applications and possible spelling errors were eliminated. Word frequency lists were obtained from the plain text files with the help of NKod software. Word frequency lists were analyzed with the help of R programming language and libraries used in automatic text analysis in RStudio software. The textbooks used in the study were accepted as four different documents (D2000, D2005, D2013 ve D2018) according to the four curricula (2000, 2005, 2013 and 2018). The research data consisted of the name frequency lists obtained from the texts in these four different documents. In this direction, data consisting of 109877 names were analyzed. In the analysis, firstly, a raw frequency list of ten terms was obtained and it was concluded that these ten terms reflect the nature of the physics subject area. Secondly, the Term Frequency-Inverse Document Frequency (TF-IDF) method, which is frequently applied in automatic text analysis, was used. This method weights the frequency of terms in the document depending on the frequency of use between documents. Thanks to this method, common words between documents and distinctive words specific to documents were identified. In the analysis of common terms, it was observed that the D2000 document differed from the other three documents, and the ratio of common words decreased with the inclusion of the D2000 document in pair and trio comparisons. The comparison of D2013 and D2018 documents was found to be the comparison with the highest common word rate. Thirdly, in the TF-IDF graph obtained, it is concluded that the D2000 document consists of physics-specific words, while words such as design, engineering, reason, problem have distinctiveness in D2005 and later documents. The reason for this may be the removal or change of objectives such as magnetism and electricity from physics units in the documents after the D2000 document. In addition, the increase in the number of objectives related to scientific process skills in D2005 and later documents can be said as another reason. This study shows that the changes that occur between programs with Turkish natural language processing and TF-IDF methods can be revealed by analyzing textbooks in science education with big data. Keywords: Textbooks, Science, Physics Education, Automated Text Analysis, Word Frequency List, TF-IDF Statistics, Big Data.