<![CDATA[Data recommendations are built by displaying the results of student subject recommendations based on students' computational thinking value. The process carried out is tokenization, stopword removal, stemming, and weighting. The extraction results were then compared using the cosine similarity approach. The greater the value of cosine similarity produced, the more similar the two data are, so that the material recommendations will be based on the smallest cosine similarity value between the extraction of student recommendation data. From the 535 data, several student data are included in 3 levels of material, namely recommendation 0 (low), recommendation 1 (medium), and recommendation 2 (high). Recommendation data was obtained from the results of students' computational thinking calculations by looking at decomposition value, pattern value, abstraction value, and algorithm value.]]>
