--- license: cc-by-sa-4.0 task_categories: - conversational - question-answering - text-generation tags: - qa - knowledge-graph language: - en dataset_info: features: - name: category dtype: string - name: size dtype: int32 - name: id dtype: string - name: eid dtype: string - name: original_triple_sets list: - name: subject dtype: string - name: property dtype: string - name: object dtype: string - name: modified_triple_sets list: - name: subject dtype: string - name: property dtype: string - name: object dtype: string - name: shape dtype: string - name: shape_type dtype: string - name: lex sequence: - name: comment dtype: string - name: lid dtype: string - name: text dtype: string - name: lang dtype: string - name: test_category dtype: string - name: dbpedia_links sequence: string - name: links sequence: string - name: graph list: list: string - name: main_entity dtype: string - name: mappings list: - name: modified dtype: string - name: readable dtype: string - name: graph dtype: string - name: dialogue list: - name: question list: - name: source dtype: string - name: text dtype: string - name: graph_query dtype: string - name: readable_query dtype: string - name: graph_answer list: string - name: readable_answer list: string - name: type list: string splits: - name: train num_bytes: 33200723 num_examples: 10016 - name: validation num_bytes: 4196972 num_examples: 1264 - name: test num_bytes: 4990595 num_examples: 1417 - name: challenge num_bytes: 420551 num_examples: 100 download_size: 9637685 dataset_size: 42808841 --- # Dataset Card for WEBNLG-QA ## Dataset Description - **Paper:** [SPARQL-to-Text Question Generation for Knowledge-Based Conversational Applications (AACL-IJCNLP 2022)](https://aclanthology.org/2022.aacl-main.11/) - **Point of Contact:** Gwénolé Lecorvé ### Dataset Summary WEBNLG-QA is a conversational question answering dataset grounded on WEBNLG. It consists in a set of question-answering dialogues (follow-up question-answer pairs) based on short paragraphs of text. Each paragraph is associated a knowledge graph (from WEBNLG). The questions are associated with SPARQL queries. ### Supported tasks The dataset can be used for: * Knowledge-based question-answering * SPARQL-to-Text conversion #### Knowledge based question-answering Below is an example of dialogue: - Q1: What is used as an instrument is Sludge Metal or in Post-metal? - A1: Singing, Synthesizer - Q2: And what about Sludge Metal in particular? - A2: Singing - Q3: Does the Year of No Light album Nord belong to this genre? - A3: Yes. #### SPARQL-to-Text Question Generation SPARQL-to-Text question generation refers to the task of converting a SPARQL query into a natural language question, eg: ```SQL SELECT (COUNT(?country) as ?answer) WHERE { ?country property:member_of resource:Europe . ?country property:population ?n . FILTER ( ?n > 10000000 ) } ``` could be converted into: ```txt How many European countries have more than 10 million inhabitants? ``` ## Dataset Structure ### Types of questions Questions are very diverses, ranging the following feature categories: * SPARQL query * Number of triplets: 1, 2, 3+ * Logical connector between clauses: conjunction, disjunction, exclusion * Topology of the query graph: direct, sibling, chain, mixed, other * Variable typing: none, target variable, internal variable * Comparison clause: none, string, number, date * Answer * Type: entity (open), entity (close list of choices), number, boolean * Number of answers: 0 (unanswerable question), 1, 2+ * Number of target variables: 0 ("ASK" verb), 1, 2 * Contextuality * Dialogue context: self-sufficient question, contains coreference, contains ellipsis * Meaning: meaninful question, non-sense question ### Data splits Text verbalization is only available for a subset of the test set, referred to as *challenge set*. Other sample only contain dialogues in the form of follow-up sparql queries. | | Train | Validation | Test | Challenge | | --------------------- | ---------- | ---------- | ---------- | ------------ | | Questions | 27727 | 3485 | 4179 | 332 | | Dialogues | 1001 | 1264 | 1417 | 100 | | NL question per query | 0 | 0 | 0 | 2 | | Characters per query | 129 (± 43) | 131 (± 45) | 122 (± 45) | 113 (± 38) | | Tokens per question | - | - | - | 8.4 (± 4.5) | ## Additional information ### Related datasets This corpus is part of a set of 5 datasets released for SPARQL-to-Text generation, namely: - Non conversational datasets - [SimpleQuestions](https://huggingface.co/datasets/OrangeInnov/simplequestions-sparqltotext) (from https://github.com/askplatypus/wikidata-simplequestions) - [ParaQA](https://huggingface.co/datasets/OrangeInnov/paraqa-sparqltotext) (from https://github.com/barshana-banerjee/ParaQA) - [LC-QuAD 2.0](https://huggingface.co/datasets/OrangeInnov/lcquad_2.0-sparqltotext) (from http://lc-quad.sda.tech/) - Conversational datasets - [CSQA](https://huggingface.co/datasets/OrangeInnov/csqa-sparqltotext) (from https://amritasaha1812.github.io/CSQA/) - [WebNLQ-QA](https://huggingface.co/datasets/OrangeInnov/webnlg-qa) (derived from https://gitlab.com/shimorina/webnlg-dataset/-/tree/master/release_v3.0) ### Licencing information * Content from original dataset: CC-BY-SA 4.0 * New content: CC BY-SA 4.0 ### Citation information ```bibtex @inproceedings{lecorve2022sparql2text, title={SPARQL-to-Text Question Generation for Knowledge-Based Conversational Applications}, author={Lecorv\'e, Gw\'enol\'e and Veyret, Morgan and Brabant, Quentin and Rojas-Barahona, Lina M.}, journal={Proceedings of the Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the International Joint Conference on Natural Language Processing (AACL-IJCNLP)}, year={2022} } ```