Update README.md

README.md

@@ -1,8 +1,12 @@
+A dataset for benchmarking keyphrase extraction and generation techniques from abstracts of english scientific articles. For more details about the dataset please refer the original paper - [https://arxiv.org/abs/1704.02853](https://arxiv.org/abs/1704.02853)
+
+Original source of the data - [https://scienceie.github.io/](https://scienceie.github.io/)
+
 ## Dataset Summary
+The Semeval-2017 dataset was originally proposed by *Isabelle Augenstein et al.* in the paper titled - [SemEval 2017 Task 10: ScienceIE - Extracting Keyphrases and Relations from Scientific Publications](https://arxiv.org/abs/1704.02853) in the year 2017. The dataset consists of a abstracts of 500 English scientific papers from the ScienceDirect open access publications. The selected articles were evenly distributed among the domains of Computer Science, Material Sciences and Physics. Each paper has a set of keyphrases annotated by student volunteers. Each paper was double-annotated, where the second annotation was done by an expert annotator. In case of disagreement, the annotations done by expert annotators were chosen. The original dataset was divided into train, dev and test splits, evenly distributed across the three domains. The train, dev and test splits had 350, 50 and 100 articles respectively.
 
-A dataset for benchmarking keyphrase extraction and generation techniques from english news articles. For more details about the dataset please refer the original paper - [https://arxiv.org/abs/1704.02853](https://arxiv.org/abs/1704.02853)
+ The dataset shared over here categorizes the keyphrases into *extractive* and *abstractive*. **Extractive keyphrases** are those that could be found in the input text and the **abstractive keyphrases** are those that are not present in the input text. In order to get all the meta-data about the documents and keyphrases please refer to the [original source](https://scienceie.github.io/) from which the dataset was taken from. The main motivation behind making this dataset available in the form as presented over here is to make it easy for the researchers to programmatically download it and evaluate their models for the tasks of keyphrase extraction and generation. As keyphrase extraction by treating it as a sequence tagging task and using contextual language models has become popular - [Keyphrase extraction from scholarly articles as sequence labeling using contextualized embeddings](https://arxiv.org/pdf/1910.08840.pdf), we have also made the token tags available in the BIO tagging format.   
 
-Original source of the data - [https://github.com/LIAAD/KeywordExtractor-Datasets/blob/master/datasets/SemEval2017.zip](https://github.com/LIAAD/KeywordExtractor-Datasets/blob/master/datasets/SemEval2017.zip)
 
 
 ## Dataset Structure