2020/10/18 by Jason Obeid, Obeid, Jason, Enamul Hoque +1 · 8 citations
Computer Science · #Data Visualization and Analytics #Video Analysis and Summarization #Handwritten Text Recognition Techniques
paper · pdf · doi:10.48550/arxiv.2010.09142
Information visualizations such as bar charts and line charts are very\npopular for exploring data and communicating insights. Interpreting and making\nsense of such visualizations can be challenging for some people, such as those\nwho are visually impaired or have low visualization literacy. In this work, we\nintroduce a new dataset and present a neural model for automatically generating\nnatural language summaries for charts. The generated summaries provide an\ninterpretation of the chart and convey the key insights found within that\nchart. Our neural model is developed by extending the state-of-the-art model\nfor the data-to-text generation task, which utilizes a transformer-based\nencoder-decoder architecture. We found that our approach outperforms the base\nmodel on a content selection metric by a wide margin (55.42% vs. 8.49%) and\ngenerates more informative, concise, and coherent summaries.\n