Integrating Importance, Non-Redundancy and Coherence in Graph-Based Extractive Summarization
Authors: Daraksha Parveen, Michael Strube
IJCAI 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We compare ROUGE scores and human judgements for coherence of different systems on scientific articles. Our method performs considerably better than other systems on this data. Also, our graph-based summarization technique achieves state-of-the-art results on DUC 2002 data. ... The datasets used, the experimental setup and the results are described in Section 4 and discussed in Section 5. |
| Researcher Affiliation | Academia | Daraksha Parveen and Michael Strube NLP Group and Research Training Group AIPHES Heidelberg Institute for Theoretical Studies g Gmb H Schloß-Wolfsbrunnenweg 35 69118 Heidelberg, Germany {daraksha.parveen michael.strube}@h-its.org |
| Pseudocode | No | The paper includes a control flow diagram (Figure 1) and mathematical equations, but no explicitly labeled 'Pseudocode' or 'Algorithm' block, nor structured code-like steps. |
| Open Source Code | No | The paper mentions using and citing several third-party software tools with their respective websites (e.g., 'http://www.smartschat.de/software/', 'http://www.ling.ohio-state.edu/ melsner/#software', 'http://www.gurobi.com'), but it does not provide any link or explicit statement about the availability of the authors' own source code for the proposed method. |
| Open Datasets | Yes | We introduce a new dataset for summarizing scientific articles. It consists of 50 articles from the high impact open access journal PLOS Medicine. ... PLOS Medicine is distributed by means of a Creative Commons Attribution License allowing us to publish the dataset. On the PLOS Medicine data our graph-based approach outperforms several baselines... Finally, we also apply our technique to the DUC 2002 data... |
| Dataset Splits | No | The paper states, 'DUC 2002 imposes a 100 word limit for the final summary. For the PLOS Medicine summaries we restrict the length in terms of the number of sentences. In the experiments reported in Section 4, we will discuss results with a 5 sentence limit for the final summary'. However, it does not provide specific details on how the datasets were partitioned into training, validation, and test sets, or specific percentages/counts for these splits. |
| Hardware Specification | No | The paper does not provide specific details regarding the hardware used to run the experiments, such as GPU models, CPU types, or memory specifications. |
| Software Dependencies | No | The paper mentions specific software tools used such as 'Stanford parser', 'coreference resolution system by [Martschat, 2013]', 'Brown coherence toolkit [Elsner and Charniak, 2011]', and 'Gurobi' (with a website link 'http://www.gurobi.com'). However, it does not provide specific version numbers for any of these software dependencies. |
| Experiment Setup | Yes | We extract the content of a paper excluding figures, table and references. ... After this we parse articles using the Stanford parser [Klein and Manning, 2003]. We perform pronoun resolution using the coreference resolution system by [Martschat, 2013]2. We apply the Brown coherence toolkit [Elsner and Charniak, 2011]3 to the articles... The importance and coherence of a sentence is used in the optimization phase4. ... DUC 2002 imposes a 100 word limit for the final summary. For the PLOS Medicine summaries we restrict the length in terms of the number of sentences. In the experiments reported in Section 4, we will discuss results with a 5 sentence limit for the final summary... |