We present TOTTO, an open-domain English table-to-text dataset with over 120,000 training examples that proposes a controlled generation task: given a Wikipedia table and a set of highlighted table cells, produce a one-sentence description. To obtain generated targets that are natural but also faithful to the source table, we introduce a dataset construction process where annotators directly revise existing candidate sentences from Wikipedia. We present systematic analyses of our dataset and annotation process as well as results achieved by several state-of-the-art baselines. While usually fluent, existing methods often hallucinate phrases that are not supported by the table, suggesting that this dataset can serve as a useful research benchmark for high-precision conditional text generation. 1 * Work done during an internship at Google.
Natural disasters such as earthquakes have profound effects on the earth's biodiversity. However, studies on immediate earthquake impacts are rarely conducted at fine scales due to logistical constraints. We conducted the first post-earthquake field survey in Wolong Nature Reserve, Wenchuan, China, less than 1 year after it was hit by a magnitude 8.0 earthquake in 2008. Since Wolong harbors approximately 10% of the endangered wild giant panda (Ailuropoda melanoleuca) population, the impact of the earthquake on the giant panda and its habitat is of particular concern. We established 15 transects in three focus areas within the Reserve where we classified occurrences of earthquake damage according to vegetation and geophysical characteristics. In the 11.2 km 2 area sampled, we recorded 156 occurrences of earthquake damage consisting of landslides and mudflows, which comprised a total area of 0.88 km 2 . Of all earthquake damage occurrences sampled, only 36% of occurences (73% of surface area) corresponded to damaged areas previously detected through broad-scale remote sensing. The remaining damaged areas mainly consisted of occurrences too small to be detected without field observation. Although there were significant losses to tree and shrub species diversity and richness in earthquake-damaged areas, remnant vegetation was found in the majority (80%) of damaged areas, suggesting the potential for forest recovery. Most earthquake-damaged areas were too steep to be classified as suitable giant panda habitat (79%). In addition, a sizable number of signs of giant panda (67) and other wildlife (148) were observed near the earthquakedamaged areas, and there appeared to be avoidance of earthquake damage only at short-range distances. This study has implications for understanding the impact of natural disasters on biodiversity and highlights the importance of fine scale on-the-ground assessments of disaster impacts on wildlife and their habitats.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.