There have been attempts in various educational settings to elucidate factors that assist the amelioration of reading comprehensions of students. Conventionally, the mainstream method of reading skills cultivation in practice has been summarization, where students are asked to shorten the length of passage by appropriately omitting non-essential words and effectively paraphrasing the words in it. Yet it poses significant challenges to educators given that, in the first place, assessing their reading skills on a quantitative basis is quite a hurdle to overcome due to highly variable patterns and qualities in submissions, which ends in hindering further analyses. In order to rectify these issues, this research sheds light upon the relatively newly introduced approach called abridgement, where students are only allowed to delete the words in the passage to construct an abstract that still holds semantic and grammatical correctness. In light of its operational property, this research employed a technique of dynamic programming so that each submission can be processed in polynomial time, besides highlighting the fact that this approach is capable of providing quantifiable and computable outputs for the following analyses. Receiving these numerical data accompanied by some comments collected from students who participated in the experiment, this research moves onto converting them into the co-occurrence networks to visualize the results so that it becomes more feasible to ascertain some of the factors that contribute to the development of their reading abilities, as well as factors that do not.