Universal Semantic Web Assistant based on Sequence to Sequence Model and Natural Language Understanding
S V Prajwal, G.N. Mamatha, P Ravi, Divya Manoj, Shri Krishna Joisa · 2019
Virtual Assistants are complex systems that try to mimic the conversation with a user through textual inputs or voice commands. Chatbots are subsets of virtual assistant and the names can be used interchangeably. Web based Virtual Assistants offer platform independency and requires a client such as, a browser to open the chatbot. Online virtual assistants or chatbots offer a lot of use-cases and can be employed in online Businesses to offer customer services and to improve customer experience. Web based virtual assistant can enable basic automations that are repetitive in nature. chatbots behave as a part of semantic search and learn through their past experiences. Chatbot can be built either as a specialist chatbot or a generalized one. Our Web based Virtual Assistant is both a generalist and a specialist in the retail and ecommerce domain. Our virtual assistant attempts to answer the users basic queries and applies efficient Natural Language Processing (NLP) techniques to extract the context of the user query. The virtual assistant scrapes the Web and queries the database through efficient web crawlers and through the Artificial Intelligence (AI), Machine Learning (ML), Deep Learning (DL) modules respectively. Each web crawlers are small programs written to crawl certain part of the web and extract useful information in unstructured data format. Efficient Machine Learning and Deep Learning techniques are applied to extract and analyze the data.