An official website of the United States government
We investigate pre-training techniques for abstractive multi-document summarization (MDS), which is much less studied than summarizing single documents. Though recent work has demonstrated the effectiveness of highlighting information salience for pretraining strategy design, they struggle to generate abstractive and reflective summaries, which are critical properties for MDS. To this end, we present PELMS, a pre-trained model that uses pre-training objectives based on semantic coherence heuristics and faithfulness constraints together with unlabeled multi-document inputs, to promote the generation of concise, fluent, and faithful summaries. To support the training of PELMS, we compile MultiPT, a multidocument pre-training corpus containing over 93 million documents to form more than 3 million unlabeled topic-centric document clusters, covering diverse genres such as product reviews, news, and general knowledge. We perform extensive evaluation of PELMS in lowshot settings on a wide range of MDS datasets. Our approach consistently outperforms competitive comparisons with respect to overall informativeness, abstractiveness, coherence, and faithfulness, and with minimal fine-tuning can match performance of language models at a much larger scale (e.g., GPT-4).
more »
« less
Do Multi-Document Summarization Models Synthesize ?
Abstract Multi-document summarization entails producing concise synopses of collections of inputs. For some applications, the synopsis should accurately synthesize inputs with respect to a key aspect, e.g., a synopsis of film reviews written about a particular movie should reflect the average critic consensus. As a more consequential example, narrative summaries that accompany biomedical systematic reviews of clinical trial results should accurately summarize the potentially conflicting results from individual trials. In this paper we ask: To what extent do modern multi-document summarization models implicitly perform this sort of synthesis? We run experiments over opinion and evidence synthesis datasets using a suite of summarization models, from fine-tuned transformers to GPT-4. We find that existing models partially perform synthesis, but imperfectly: Even the best performing models are over-sensitive to changes in input ordering and under-sensitive to changes in input compositions (e.g., ratio of positive to negative reviews). We propose a simple, general, effective method for improving model synthesis capabilities by generating an explicitly diverse set of candidate outputs, and then selecting from these the string best aligned with the expected aggregate measure for the inputs, or abstaining when the model produces no good candidate.
more »
« less
- Award ID(s):
- 2211954
- PAR ID:
- 10593405
- Editor(s):
- Louis, Annie
- Publisher / Repository:
- Association for Computational Linguistics
- Date Published:
- Journal Name:
- Transactions of the Association for Computational Linguistics
- Volume:
- 12
- ISSN:
- 2307-387X
- Page Range / eLocation ID:
- 1043 to 1062
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Many corpora span broad periods of time. Language processing models trained during one time period may not work well in future time periods, and the best model may depend on specific times of year (e.g., people might describe hotels differently in reviews during the winter versus the summer). This study investigates how document classifiers trained on documents from certain time intervals perform on documents from other time intervals, considering both seasonal intervals (intervals that repeat across years, e.g., winter) and non-seasonal intervals (e.g., specific years). We show experimentally that classification performance varies over time, and that performance can be improved by using a standard domain adaptation approach to adjust for changes in time.more » « less
-
Proc. 2023 The Web Conf. (Ed.)Summarizing text-rich documents has been long studied in the literature, but most of the existing efforts have been made to summarize a static and predefined multi-document set. With the rapid development of online platforms for generating and distributing text-rich documents, there arises an urgent need for continuously summarizing dynamically evolving multi-document sets where the composition of documents and sets is changing over time. This is especially challenging as the summarization should be not only effective in incorporating relevant, novel, and distinctive information from each concurrent multi-document set, but also efficient in serving online applications. In this work, we propose a new summarization problem, Evolving Multi-Document sets stream Summarization (EMDS), and introduce a novel unsupervised algorithm PDSum with the idea of prototype-driven continuous summarization. PDSum builds a lightweight prototype of each multi-document set and exploits it to adapt to new documents while preserving accumulated knowledge from previous documents. To update new summaries, the most representative sentences for each multi-document set are extracted by measuring their similarities to the prototypes. A thorough evaluation with real multi-document sets streams demonstrates that PDSum outperforms state-of-the-art unsupervised multi-document summarization algorithms in EMDS in terms of relevance, novelty, and distinctiveness and is also robust to various evaluation settings.more » « less
-
Extractive text summarization aims at extract- ing the most representative sentences from a given document as its summary. To extract a good summary from a long text document, sen- tence embedding plays an important role. Re- cent studies have leveraged graph neural net- works to capture the inter-sentential relation- ship (e.g., the discourse graph) to learn con- textual sentence embedding. However, those approaches neither consider multiple types of inter-sentential relationships (e.g., semantic similarity & natural connection), nor model intra-sentential relationships (e.g, semantic & syntactic relationship among words). To ad- dress these problems, we propose a novel Mul- tiplex Graph Convolutional Network (Multi- GCN) to jointly model different types of rela- tionships among sentences and words. Based on Multi-GCN, we propose a Multiplex Graph Summarization (Multi-GraS) model for extrac- tive text summarization. Finally, we evaluate the proposed models on the CNN/DailyMail benchmark dataset to demonstrate the effec- tiveness of our method.more » « less
-
We provide an overview of the MSLR2022 shared task on multi-document summarization for literature reviews. The shared task was hosted at the Third Scholarly Document Processing (SDP) Workshop at COLING 2022. For this task, we provided data consisting of gold summaries extracted from review papers along with the groups of input abstracts that were synthesized into these summaries, split into two subtasks. In total, six teams participated, making 10 public submissions, 6 to the Cochrane subtask and 4 to the MSˆ2 subtask. The top scoring systems reported over 2 points ROUGE-L improvement on the Cochrane subtask, though performance improvements are not consistently reported across all automated evaluation metrics; qualitative examination of the results also suggests the inadequacy of current evaluation metrics for capturing factuality and consistency on this task. Significant work is needed to improve system performance, and more importantly, to develop better methods for automatically evaluating performance on this task.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.1162/tacl_a_00687
