Category: Semantic technologies (Page 13 of 72)

Google introduced “ToTTo: A Controlled Table-To-Text Generation Dataset”, an open domain table-to-text generation dataset created using a novel annotation process (via sentence revision) along with a controlled text generation task that can be used to assess model hallucination. ToTTo (shorthand for “Table-To-Text”) consists of 121,000 training examples, along with 7,500 examples each for development and test. Due to the accuracy of annotations, this dataset is suitable as a challenging benchmark for research in high precision text generation. The dataset and code are open-sourced on our GitHub repo.

In the last few years, research in natural language generation, used for tasks like text summarization, has made tremendous progress. Yet, despite achieving high levels of fluency, neural systems can still be prone to hallucination (i.e.generating text that is understandable, but not faithful to the source), which can prohibit these systems from being used in many applications that require high degrees of accuracy.

While the process of assessing the faithfulness of generated text to the source content can be challenging, it is often easier when the source content is structured (e.g., in tabular format). Moreover, structured data can also test a model’s ability for reasoning and numerical inference. However, existing large scale structured datasets are often noisy (i.e., the reference sentence cannot be fully inferred from the tabular data), making them unreliable for the measurement of hallucination in model development.

https://ai.googleblog.com/2021/01/totto-controlled-table-to-text.html

AI inside Inc. announced a new service to support multiple languages on AI-OCR “DX Suite”, through the release of new AI-engine that can recognize English, Traditional Chinese, Thai, and Vietnamese characters. Through this AI inside will begin its global expansion starting with the Asian market. Through the provision of AI-OCR “DX Suite”, AI inside has contributed to improve the operational efficiency and productivity of companies and municipalities in Japan through the high accuracy recognition of both printed and handwritten characters. AI inside has also been developing a foreign language recognition AI-engine in order to expand the availability of “DX Suite” beyond Japan to other countries around the world. This foreign language AI-engine has now achieved commercially viable accuracy to recognize characters of English, Traditional Chinese, Thai, and Vietnamese, and now announce the availability of this multilingual service on the cloud version of “DX Suite”. For current users of “DX Suite” cloud version it is possible to utilize this multilingual service without any additional registration.

https://dx-suite.com/global/lp/