Content selection mainly determines what content to select from the input table, whereas surface realization primarily generates text from the selected content. In general, table-to-text generation is divided into two subtasks: content selection and surface realization. It has been applied in various scenarios, such as biography generation, NBA game generation, and weather forecasting 1– 3. Natural language generation based on tabular data, also known as table-to-text generation, takes tabular data as input and generates human-like expressions of text. On ROTOWIRE, the result is increased by 4.29% on CO metric, and 1.93 points higher on BLEU. On WIKIBIO, the result is improved from 45.47 to 46.87 on BLEU and from 41.54 to 42.28 on ROUGE. The model is verified by experiments on two datasets and better results are obtained than the baseline model. Furthermore, we improve the text search strategy to reduce the probability of generating incoherent and repetitive sentences. Table-text constraint loss is used to effectively model table inputs, whereas copy loss is exploited to precisely copy word fragments from a table. Then we propose two auxiliary learning objectives, namely table-text constraint loss and copy loss. Firstly, to make the model better learn the semantic relevance between table and text, we apply a word transformation method, which incorporates the field and position information into the target text to acquire the position of where to copy. In order to overcome this problem, we invent an auto-regressive framework based on the transformer that combines a copying mechanism and language modeling to generate target texts. Copying words from the table is a common method to solve the “out-of-vocabulary” problem, but it’s difficult to achieve accurate copying. Generating fluent, coherent, and informative text from structured data is called table-to-text generation.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |