ArxivDIGESTables: Synthesizing Scientific Literature into Tables using Language Models

When conducting literature reviews, scientists often create literature review tablestables whose rows are publications and whose columns constitute a schema, a set of aspects used to compare and contrast the papers. Can we automatically generate these tables using language models (LMs)? In this work, we introduce a framework that leverages LMs to perform this task by decomposing it into separate schema and value generation steps. To enable experimentation, we address two main challenges: First, we overcome a lack of high-quality datasets to benchmark table generation by curating and releasing ARXIVDIGESTABLES, a new dataset of 2,228 literature review tables extracted from ArXiv papers that synthesize a total of 7,542 research papers. Second, to support scalable evaluation of model generations against humanauthored reference tables, we develop DECON-TEXTEVAL, an automatic evaluation method that aligns elements of tables with the same underlying aspects despite differing surface forms. Given these tools, we evaluate LMs' abilities to reconstruct reference tables, finding this task benefits from additional context to ground the generation (e.g. table captions, in-text references). Finally, through a human evaluation study we find that even when LMs fail to fully reconstruct a reference table, their generated novel aspects can still be useful. blnewman/arxivDIGESTables bnewm0609/arxivDIGESTables * Equal contributions. Dataset size Annotation method Intended Application Evaluation Metric Paper 1 1,200 video sequences Subjectively annotated Objective VQA method development Subjective Mean Opinion Score Paper 2 585 videos Subjective video quality scores via crowdsourcing NR video quality prediction advancement Subjective video quality scores Paper 3 153,841 videos Coarsely annotated set with five quality ratings each Deep-learning VQA model training Spearman rank-order correlation coefficient Paper 4 1 million YouTube videos N/A Large-scale video classification and action recognition Performance improvements over baselines Dataset Size Task Annotations