Past Written Texts Dataset

The dataset consists of features extracted from older adults’ text.

The texts were written by the older person either in an electronic mean (eg. older e-mail), or in paper form and were transcribed by the project's clinical nurses.

The texts were then translated to English using the MyMemory service (https://mymemory.translated.net/), and a series of features were generated that can be used for sentiment analysis.

The list of fields of this dataset is presented below:

- <strong>Part_id</strong>: The user ID, which should be a 4-digit number

- <strong>Date</strong>: The recording date, which follows the “DD-MM-YY” format (eg. 14 September 2017, is formatted as 14-09-17)

- <strong>Clinical_visit</strong>: As several clinical evaluations were performed to each older adult, this number shows for which clinical evaluation these measurements refer to

- <strong>Transcript</strong>: If the text was written by the older adult (0) or was transcribed by a nurse (1)

- <strong>Language</strong>: The original language of the text (0 = Greek)

- <strong>Text_length, Number_of_sentences, Number_of_words, Number_of_words_per_sentence, Text_entropy</strong>: Statistical Measures

- <strong>Desc_image_ENG_sentiment, Desc_event_sentiment, Prev_text_ENG_sentiment</strong>: Sentiment Analysis

- <strong>Tf-XX</strong>: Term frequency – Inverse document frequency

- <strong>Tf-pos-XX</strong>: Part of Speech analysis, using tf-idf methodology