Automatic Scientific Annotation

This repository contains the automatically annotated citations for the ScisummNet dataset using the classifier trained on the annotated collection scisumm-corpus. The annotation method uses the pivot-based classification model presented in this paper.

Repository Organization

The repository is organized as follow:

• The pivot_words directory contains the pivot words both for the citing and the cited sentences extracted from the scisumm-corpus training set using the method described here. Implementation available in this repository.
• The annotations folder contains the output of the automatic annotation described in the paper paper. For each paper the annotations are proposed in a .tsv where each line is formatted as: citation_id citation_type.
• The hyp_facet_summaries directory contains one subfolder for each paper in the test set. Each subfolder contains the proposed facet summaries by different regression models.
• The ref_facet_summaries directory contains the gold facet summaries for each pair paper - citation facet.
• The reproducibility directory contains the train/test key split to reproduce the results of the paper.

Summarization Examples

Facet	Type	Text
Method	Citing	We used the base feature model defined in (Nivre et al, 2006) for all the languages but Arabic, Chinese, Czech, and Turkish.
	Cited	Features of the type DEPREL have a special status in that they are extracted during parsing from the partially built dependency graph and may therefore contain errors, whereas all the other features have gold standard values during both training and parsing.2 Based on previous research, we defined a base model to be used as a starting point for languagespecific feature selection.
Results	Citing	These parameters are identical to Nivre et al (2006b) to enable a,comparison of the scores. We evaluated the feature candidates on a,development set using the labeled and unlabeled attachment scores (LAS,and UAS) that we computed with the eval.pl script from CoNLL-X.
	Cited	This is noticeable for German (Brants et al., 2002) and Portuguese,(Afonso et al., 2002), which still have high overall accuracy thanks to,very high attachment scores, but much more conspicuous for Czech,(B¨ohmov´a et al., 2003), Dutch (van der Beek et al., 2002) and Slovene,(Dˇzeroski et al., 2006), where root precision drops more drastically to,about 69%, 71% and 41%, respectively, and root recall is also affected,negatively.
Implication	Citing	We split German compound words (mostly nouns), based on the frequency of the words in the potential decompositions (Koehn and Knight, 2003a).
	Cited	One reason for this is that the system recovers more easily from words that are split too much than from words that are not split up sufficiently.
Aim	Citing	the possible part(s)-of-speech of unknown words (Mikheev,1997).
	Cited	In this paper we present a technique for fully automatic acquisition of rules that guess possible part-of-speech tags for unknown words using their starting and ending segments.

Citation

Access to publication: https://rdcu.be/b5rel

The material is available for research purposes only. If you use this material in your research work please cite:

La Quatra, M., Cagliero, L. & Baralis, E. Exploiting pivot words to classify and summarize discourse facets of scientific papers. Scientometrics (2020). https://doi.org/10.1007/s11192-020-03532-3

@article{10.1007/s11192-020-03532-3, 
  author = {Quatra, Moreno La and Cagliero, Luca and Baralis, Elena}, 
  title = {{Exploiting pivot words to classify and summarize discourse facets of scientific papers}}, 
  issn = {0138-9130}, 
  doi = {10.1007/s11192-020-03532-3}, 
  abstract = {{The ever-increasing number of published scientific articles has prompted the need for automated, data-driven approaches to summarizing the content of scientific articles. The Computational Linguistics Scientific Document Summarization Shared Task (CL-SciSumm 2019) has recently fostered the study and development of new text mining and machine learning solutions to the summarization problem customized to the academic domain. In CL-SciSumm, a Reference Paper (RP) is associated with a set of Citing Papers (CPs), all containing citations to the RP. In each CP, the text spans (i.e., citances) have been identified that pertain to a particular citation to the RP. The task of identifying the spans of text in the RP that most accurately reflect the citance is addressed using supervised approaches. This paper proposes a new, more effective solution to the CL-SciSumm discourse facet classification task, which entails identifying for each cited text span what facet of the paper it belongs to from a predefined set of facets. It proposes also to extend the set of traditional CL-SciSumm tasks with a new one, namely the discourse facet summarization task. The idea behind is to extract facet-specific descriptions of each RP consisting of a fixed-length collection of RP’s text spans. To tackle both the standard and the new tasks, we propose machine learning supported solutions based on the extraction of a selection of discriminating words, called pivot words. Predictive features based on pivot words are shown to be of great importance to rate the pertinence and relevance of a text span to a given facet. The newly proposed facet classification method performs significantly better than the best performing CL-SciSumm 2019 participant (i.e., the classification accuracy has increased by + 8\%), whereas regression methods achieved promising results for the newly proposed summarization task.}}, 
  pages = {1--19}, 
  journal = {Scientometrics}, 
  year = {2020}
}