This repository behaves as a support material for the paper entitled "Personalizable Virtual Assistants for Question Answering over Knowledge Graphs" submitted (and under evaluation) as a resource paper at ESWC 2021. It provides a Permanent URI for the Virtual Assistant Generator (virtual_assistant_generator.zip) that, at its actual version v 1.0, enables the automatic generation of Alexa skills to reply to natural language questions by querying SPARQL endpoints. The operating mechanism of the generator is detailed in Section 5 of the article. To demonstrate the operating mechanisms of the generator, we create the following Alexa skills as use cases, documented in Section 6 of the paper: - Wikidata_usecase.zip contains the two versions of the Wikidata Alexa skill to show the possibility to customize the linking mode performed by the generator. While in the dictionary-based WikiSkill we exploit a dictionary to perform the entity and relation linking, in the API-driven WikiSkill, we perform the linking phase by relying on the Wikidata APIs. For each skill, we report 1) the Alexa skill components (back_end.zip that implements the skill logic in Node.js and the interaction_model.js that defines the skill intents); 2) the conf.json that corresponds to the configuration file to generate the Alexa skill by our generator (when applicable); 3) the replies corresponding to the evaluation of the Wikidata skills on the QALD-7 training set; - DBpedia_usecase.zip contains the DBpedia Alexa skills. They are used to demonstrate the different generator configuration options: users can manually create the generator configuration file or auto-generate them, and use it as-is or manually refine it before generating the Alexa skill. For each DBpedia skill, we report 1) the Alexa skills components (back_end.zip that implements the skill logic in Node.js and the interaction_model.js that defines the skill intents); 2) the conf.json that corresponds to the configuration file to generate the Alexa skill by our generator; 3) its evaluation on the QALD-9 training set; - UNESCO_personal_assistant.zip contains 1) the UNESCO Thesaurus Alexa skills component (back_end.zip that implements the skill logic in Node.js and the interaction_model.js that defines the skill intents); 2) the conf.json that corresponds to the configuration file to generate the Alexa skill by our generator; - wordnet_personal_assistant.zip contains 1) the WordNet Alexa skills component (back_end.zip that implements the skill logic in Node.js and the interaction_model.js that defines the skill intents); 2) the conf.json that corresponds to the configuration file to generate the Alexa skill by our generator; - warsampo_personal_assistant.zip contains 1) the WarSampo Alexa skills component (back_end.zip that implements the skill logic in Node.js and the interaction_model.js that defines the skill intents); 2) the conf.json that corresponds to the configuration file to generate the Alexa skill by our generator; The content of this repository is also provided on GitHub at https://github.com/mariaangelapellegrino/virtual_assistant_generator. The entire project is released under the MIT license.

This repository behaves as a support material for the paper entitled "Personalizable Virtual Assistants for Question Answering over Knowledge Graphs" submitted (and under evaluation) as resource paper at ESWC 2021. The content of this repository is also provided on GitHub at https://github.com/mariaangelapellegrino/virtual_assistant_generator.