Examining the State-of-the-Art in News Timeline Summarization

  title={Examining the State-of-the-Art in News Timeline Summarization},
  author={Demian Gholipour Ghalandari and Georgiana Ifrim},
Previous work on automatic news timeline summarization (TLS) leaves an unclear picture about how this task can generally be approached and how well it is currently solved. This is mostly due to the focus on individual subtasks, such as date selection and date summarization, and to the previous lack of appropriate evaluation metrics for the full TLS task. In this paper, we compare different TLS strategies using appropriate evaluation frameworks, and propose a simple and effective combination of… Expand

Figures, Tables, and Topics from this paper

Extractive Timeline Summarization based on Unsupervised Techniques
With the increasing importance of internet during time, a huge amount of news article about several arguments are published in several websites: users that are interested in a particular event couldExpand
Automated Timeline Length Selection for Flexible Timeline Summarization
A better TLS approach for automatically and dynamically determining the TLS timeline length is presented by employing the established elbow method from the machine learning community to automatically find the minimum number of dates within the time series to generate concise and informative summaries. Expand
Summarize Dates First: A Paradigm Shift in Timeline Summarization
This paper presents a new approach, namely Summarize Date First, which focuses on first generating date-level summaries then selecting the most relevant dates on top of summarized knowledge, which is superior to state-of-the-art unsupervised methods and competitive against supervised ones. Expand
A Proposal: Interactively Learning to Summarise Timelines by Reinforcement Learning
Timeline Summarisation (TLS) aims to generate a concise, time-ordered list of events described in sources such as news articles. However, current systems do not provide an adequate way to adapt toExpand
Event Occurrence Date Estimation based on Multivariate Time Series Analysis over Temporal Document Collections
This paper proposes an approach to temporal profiling of event mentions in text that utilizes a news article archival collection for collecting temporal as well as textual information containing contemporary and retrospective event references and outperforms the existing methods by a large margin at all granularities. Expand
TopicRefine: Joint Topic Prediction and Dialogue Response Generation for Multi-turn End-to-End Dialogue System
  • Hongru Wang, Mingyu Cui, Zimo Zhou, G. Fung, Kam-Fai Wong
  • Computer Science
  • 2021
A multi-turn dialogue always follows a specific topic thread, and topic shift at the discourse level occurs naturally as the conversation progresses, necessitating the model’s ability to captureExpand


Leveraging Learning To Rank in an Optimization Framework for Timeline Summarization
This paper proposes an optimization framework and demonstrates the use of Learning To Rank (LTR) to automatically construct timeline summaries from Web news articles, showing that this approach outperforms existing solutions in producing high quality Timeline summaries. Expand
Timeline Summarization from Relevant Headlines
This work presents a new approach that exploits the headlines of online news articles instead of the articles’ full text and outperforms state-of-the-art system in terms of relevance and understandability. Expand
A Temporally Sensitive Submodularity Framework for Timeline Summarization
It is shown that MDS optimization models using submodular functions can be adapted to yield well-performing TLS models by designing objective functions and constraints that model the temporal dimension inherent in TLS. Expand
Timeline Generation through Evolutionary Trans-Temporal Summarization
Evaluation results in ROUGE metrics indicate the effectiveness of the proposed approach based on trans-temporal information, and a novel combination of inter-date and intra-date sentence dependencies is presented. Expand
Abstractive Timeline Summarization
This work suggests an abstractive timeline summarization system that is entirely unsupervised, which makes it especially suited to TLS where there are very few gold summaries available for training of supervised systems. Expand
Revisiting the Centroid-based Method: A Strong Baseline for Multi-Document Summarization
This paper applies the centroid-based model for extractive document summarization to possible summaries instead of sentences and uses a simple greedy algorithm to find the best summary, serving a higher performance over the orig- inal model on par with more complex state-of-the-art methods. Expand
Evolutionary Hierarchical Dirichlet Process for Timeline Summarization
A novel model called Evolutionary Hierarchical Dirichlet Process (EHDP) is developed to capture the topic evolution pattern in timeline summarization and demonstrates the good performance of the system with regard to ROUGE scores. Expand
Improving ROUGE for Timeline Summarization
Variants of ROUGE that allow alignment of daily summaries via temporal distance or semantic similarity are introduced and the suitability of these variants in a theoretical analysis and in a battery of task-specific tests is argued for. Expand
A Low-Rank Approximation Approach to Learning Joint Embeddings of News Stories and Images for Timeline Summarization
This work leverages the success of matrix factorization techniques from recommender systems, and cast the problem as a sentence recommendation task, using a representation learning approach, to augment text-only corpora and propose a scalable low-rank approximation approach for learning joint embeddings of news stories and images. Expand
Evolutionary timeline summarization: a balanced optimization framework via iterative substitution
A novel framework for the web mining problem named Evolutionary Timeline Summarization (ETS), which aims to return the evolution trajectory along the timeline, consisting of individual but correlated summaries of each date, emphasizing relevance, coverage, coherence and cross-date diversity. Expand