QAsystems（斯坦福问答系统）.pdf

QAsystems （斯坦福问答系统） 最近在阅读⼀些AI项⽬，写⼊markdown，持续更新，算是之后也能回想起做法 QA systems(问答系统) tutorial(指导)： Question answering is an important NLP task and longstanding milestone for artificial intelligence systems. QA systems allow a user to ask a question in natural language, and receive the answer to their question quickly and succinctly.(问答是⼀ 项重要的NLP任务，是⼈⼯智能的长久基⽯。问答系统让⽤户⽤⾃然语⾔提问，然后能快速有效地给予回复) The ability to read a piece of text and then answer questions about it is called reading comprehension. Reading comprehension is challenging for machines, requiring both understanding of natural language and knowledge about the world.(阅读部分⽂本然后回答的能⼒称为阅读理解。阅读理解对于机器⽽⾔是⼀⼤挑战，不仅需要对⾃然语⾔的理解，还需要来⾃于世界 的知识) SQuAD Dataset(数据集) Stanford Question Answering Dataset (SQuAD) is a new reading comprehension dataset, consisting of questions posed by crowdworkers on a set of Wikipedia articles, where the answer to every question is a segment of text, or span, from the corresponding reading passage. With 100,000+ question-answer pairs on 500+ articles, SQuAD is significantly larger than previous reading comprehension datasets.(斯坦福问答数据集是⼀个新的阅读理解数据集，由群体⼯作者在⼀组维基百科⽂章中提出 的问题，其中每个问题的答案是⼀段⽂本或跨度，从相应的阅读段落。100000 +问答对500 +⽂章，斯坦福问答数据集明显⼤于以前的阅 读理解数据集。) Problem(问题) For each observation in the training set, we have a context, question, and text.(对于训练集的每⼀次观察，我们有内容、问题和⽂ 本) The goal is to find the text for any new question and context provided. This is a closed dataset meaning that the answer to a question is always a part of the context and also a continuous span of context. I have broken this problem into two parts for now(⽬标是在任意新问题和提供的内容中找到对应⽂本。这是⼀个相关的数据集，回复往往是内容的⼀部分和⼀段连续的跨度。我将这个 问题分为两个部分) - 1、Getting the sentence having the right answer (highlighted yellow)(获取含有正确答案的内容(图中黄⾊标注)) 2、Once the sentence is finalized, getting the correct answer from the sentence (highlighted green)(⼀旦内容准备好了，获取 其正确答案(图中绿⾊标注)) Introducing Infersent, Facebook Sentence Embedding(介绍句⼦嵌⼊) These days