LULCC-KnowText - annotated text segments for knowledge extraction on Land Use and Land Cover change
Domain:
environment and energynatural language processingRecord type:
dataset
Creator:
QuiCreValTei
Editor:
co coVal
Publisher:
CIR
Host:
This dataset contains a corpus of annotated text segments (sentences) extracted from scientific articles on Land Use and Land Cover change (LULCC) in West Africa. Each segment has been evaluated for relevance and categorized according to themes such as LULC, LULCC, DRIVERS, PROCESS, and PRACTICES, supporting automatic knowledge extraction and machine learning applications.
The dataset consists of 8 files:
- Annotated sentences, divided into 4 tabular files:
- annotated_sentences_final_corpus.csv: the whole annotated corpus
- annotated_sentences_three_annot_corpus.csv: a subset of sentences annotated by three annotators
- annotated_sentences_two_annot_corpus.csv: a subset of sentences annotated by two annotators
- annotated_sentences_single_annot_corpus.csv: a subset of sentences annotated by one annotator
The sets annotated by three annotators and two annotators were used to evaluate inter-annotator agreement during the refining of annotation guidelines.
- datasets_columns_description.pdf: This file contains the description of column names used in each dataset.
- metadata_articles.pdf: This file contains the metadata of the scientific articles from which the sentences were extracted for annotation.
- sentences_annotation_guidelines.pdf: The annotation guidelines used to manually annotate the sentences.