Improving academic writing is a key towards successful publishing, and assessment of pre-published write-ups is the beginning of that ending. Words have 'emotions', and the ability to control it in write-ups is an evolving industry skill. This paper aimed to profile and analyze through web-based sentiment analysis and text mining with theme generation of the scientific paper abstracts of junior researchers in the Philippines. Significant findings showed that majority of the specimen abstracts conformed with academic writing standards, and obtained a passing rating from conference judges. Sentiment Analysis revealed an equal writing behavior of junior researchers in wording abstracts positively or negatively, even if majority were found to have written abstracts neutrally. Text Mining reported the frequent use of the words 'effective' and 'potential' for positively-worded and negatively-worded conclusions of abstracts, respectively. Generating themes from these premises, thematic analysis proposed four principles of writing abstracts to attract acceptance of papers in science fairs. 'Sentheme Mining' as a research procedure was recommended, among others things, for use in future studies involving qualitative studies of paper abstracts or full articles.