Inappropriate Personal Reference
Academic writing is focused on accurate, reliable and objective information rather than on who conducted the research. This means the pronoun “I” is not often used, even in the discipline of computer science. The pronoun “we”, however, can be used in some academic disciplines. It is not only employed to refer to a co-authored text but also to a single-authored text.
Similarly, to maintain impersonal objectivity, writers should avoid addressing the reader directly with the use of “you” or “let’s”. The use of the possessive pronoun ‘my’ is also not advised as it is a subjective reference.
The left side of the table below highlights inappropriate use of personal pronouns. A more academically acceptable version of the text is given on the right side of the table.
Examples of personal reference
The raw Twitter text data were lightly preprocessed and the features were extracted by regular expression patterns. I did two sets of experiments. In the first set, I have classified tweets in each topic using different feature sets. The classifier is SVM with a linear kernel. Since SVM inherently applies binary classification, the multi-class case is handled by the one-vs-all paradigm. In the second set, I have applied the best feature set from my previous results to three datasets at different levels. For all classification tasks, I have reported the F1 (the harmonic mean of precision and recall) scores from ten-fold cross validation.
Academically-appropriate equivalent
The raw Twitter text data were lightly preprocessed and the features were extracted by regular expression patterns. Two sets of experiments were conducted. In the first set, tweets were classified in each topic using different feature sets. The classifier is SVM with a linear kernel. Since SVM inherently applies binary classification, the multi-class case is handled by the one-vs-all paradigm. In the second set, the most reliable feature set from the previous results was applied to three datasets at different levels. For all classification tasks, the F1 (the harmonic mean of precision and recall) scores were reported from ten-fold cross validation.