Decontextualization: Making Sentences Stand-Alone

Abstract

Models for question answering, dialogue agents, and summarization often interpret the meaning of a sentence in a rich context and use that meaning in a new context. Taking excerpts of text can be problematic, as key pieces may not be explicit in a local window. We isolate and define the problem of sentence decontextualization: taking a sentence together with its context and rewriting it to be interpretable out of context, while preserving its meaning. We describe an annotation procedure, collect data on the Wikipedia corpus, and use the data to train models to automatically decontextualize sentences. We present preliminary studies that show the value of sentence decontextualization in a user-facing task, and as preprocessing for systems that perform document understanding. We argue that decontextualization is an important subtask in many downstream applications, and that the definitions and resources provided can benefit tasks that operate on sentences that occur in a richer context.

CONQRR: Conversational Query Rewriting for Retrieval with Reinforcement Learning

“…Conversational QR is initially proposed to improve a model's understanding of the dialogue context (Elgohary et al, 2019), and gets recently adopted to address more downstream tasks like conversational retrieval and question answering (Anantha et al, 2021;Dalton et al, 2020;Yu et al, 2020). Query rewriting is also related to concepts like sentence de-contextualization (Choi et al, 2021) and prompt-based learning (Liu et al, 2021).…”

Can NLI Models Verify QA Systems' Predictions?

“…Prior work has paved the way for this application of NLI. Pieces of this pipeline like converting a question to a declarative sentence (Demszky et al, 2018) and reformulating an answer sentence to stand on its own (Choi et al, 2021) have been explored in recent work. Moreover, results from other domains such as fact-checking (Kryscinski et al, 2020;Mishra et al, 2020) and summary reranking (Falke et al, 2019) show the potential for broader applicability of NLI.…”

“…Mapping QA to NLI enables us to exploit both NLI and QA datasets for answer verification, but it relies on a pipeline for mapping (question, answer, context) triplet to a (premise, hypothesis) NLI pair. We implement a strong pipeline here: we extract a concise yet sufficient premise through decontextualization (Choi et al, 2021), which rewrites a single sentence from a document such that it can retain the semantics when presented alone without the document. We improve a prior question conversion model (Demszky et al, 2018) with a stronger pretrained seq2seq model (Raffel et al, 2020).…”

“…Richardson et al (2013) explore a similar pipeline for answering multiple-choice questions in the MCTest dataset, but find that it works quite poorly, possibly due to the low performance of RTE systems at the time (Stern and Dagan, 2011). We believe that a combination of recent advances in natural language generation (Demszky et al, 2018;Choi et al, 2021) and strong models for NLI (Liu et al, 2019) give us tools to significantly advance this approach.…”

“…We tackle this problem by decontextualizing the sentence containing the answer from the full context to make it stand alone. Recent work (Choi et al, 2021) proposed a sentence decontextualization task in which a sentence together with its context are taken and the sentence is rewritten to be interpretable out of context if feasible, while preserving its meaning. This procedure can involve name completion (e.g., Stewart → Kristen Stewart), noun phrase/pronoun swap, bridging anaphora resolution, and more.…”