Analysis of secondary school science textbooks and auxiliary resource books with text mining: The case of physical events subject area

The aim of this study is to analyse the Ministry of National Education (MoNE) secondary school science textbooks (ST) and auxiliary resource books (ARB) with lectures in the subject area of physical events by text analysis. For the analysis of the sub-problems determined in line with this purpose, text files of the books were created and analyses were made using natural language processing (NLP) and text mining methods. In the natural language processing process, word frequency lists of the relevant text files were reached with the help of NKod software prepared in JAVA programming language using Zemberek NLP Application Programming Interface. The files belonging to the word frequency lists were transferred to the R Studio programme and the necessary analyses were made using various packages. In the analysis of the first sub-problem, the first 20 words were analysed from the rooted word frequency lists of the STs and ARBs. In the analysis of the second sub-problem, a chi-square test was conducted to find the words that statistically differed between the frequencies of noun-rooted words belonging to STs and ARBs. In the analysis of the third sub-problem, the similarity between the documents belonging to STs and ARBs was examined. For this purpose, cosine similarity, which is frequently used in data analysis, was calculated and document clustering was performed. When six clusters were formed among the documents, quality clusters were reached. In addition to this analysis, similarity values were calculated between the frequency lists of noun-rooted word frequency lists of STs and ARBs in terms of book, class and unit level. As a result, it was determined that there was no differentiation in the first 10 words in the frequency lists of noun-rooted word frequency lists of the books, the differentiation was in the words with less frequency, and according to the chi-square test, it was determined that the STs generally differentiated in terms of activity/experiment materials and the ARBs differentiated in terms of concepts. In the clustering study, when the cosine similarity values of the documents were analysed, it was determined that the similarity between the STs and the ARBs was higher than the similarity between the STs. Similarity values of 0.956, 0.89 at the 5th grade level, 0.904 at the 6th grade level, 0.918 at the 7th grade level, and 0.935 at the 8th grade level were found in all of the contents belonging to the subject area of physical events. The fact that the similarity values between the STs and the ARBs are high shows that the books are not very different in terms of the frequencies of words with noun roots.