Adapting Graph Summaries to the Users’ Reading Levels

Abstract

Deciding on the complexity of a generated text in NLG systems is a contentious task. Some systems propose the generation of simple text for low-skilled readers; some choose what they anticipate to be a "good measure" of complexity by balancing sentence length and number of sentences (using scales such as the D-level sentence complexity) for the text; while others target high-skilled readers. In this work, we discuss an approach that aims to leverage the experience of the reader when reading generated text by matching the syntactic complexity of the generated text to the reading level of the surrounding text. We propose an approach for sentence aggregation and lexical choice that allows generated summaries of line graphs in multimodal articles available online to match the reading level of the text of the article in which the graphs appear. The technique is developed in the context of the SIGHT (Summarizing Information Graphics Textually) system. This paper tackles the micro planning phase of sentence generation discussing additionally the steps of lexical choice, and pronominalization.

Generating Summaries of Line Graphs

“…An evaluation of the text summaries generated at different reading level is presented at (P. Moraes et al, 2014). It shows that, indeed, different users have different preferences regarding different text designs.…”

“…Much of them use machine learning techniques in order to learn text constructions and lexicalization used in different grade levels. As presented in (P. Moraes et al, 2014), simpler and well established reading level measurement techniques suffice for our scenario. The work shows that Flesh-Kincaid (Kincaid, Fishburne, Rogers, & Chissom, 1975) and SMOG (Laughlin, 1969) provide the set of information needed by the system in order to make decisions of syntactical text complexity.…”

“…Finally, the conclusion section of the summary presents computational information about the graphic (overall value and rate change, time span of the graphic, maximum and minimum points and dates when they occur). The strategies on organizing the summaries are described in (P. Moraes, McCoy, & Carberry, 2014).…”

“…From these one can see the different aggregation and lexical choice decisions made for the different reading levels. The system also includes appropriate pronominalization in order to avoid repetition of the referring expressions (P. Moraes et al, 2014). For the surface realization phase we use FUF/SURGE (Elhadad & Robin, 1999) to create the templates for realization.…”

Towards Generating Personalized Hospitalization Summaries

“…There are several existing systems that produce personalized content in biomedical domain (Jimison et al, 1992;DiMarco et al, 1995) as well as in non-medical domains (Paris, 1988;Moraes et al, 2014). However, only a few of the existing biomedical systems generate personalized content for the patients (Buchanan et al, 1995;Williams et al, 2007).…”