Google today unveiled Data QnA, a service in private alpha that’s designed to help users get answers to analytical queries through natural language questions. Based on the Analyza system developed at Google Research, Data QnA automatically performs semantic parsing and mapping to data sets up to petabytes in size, and it can be embedded in chatbots, spreadsheets, and business intelligence platforms like Looker as well as custom-built interfaces.
Data QnA ostensibly solves the problem of business data requests. Enterprise employees typically have to ask business intelligence teams for dashboards and reports, a process that can take days and yield abstracts that fail to answer follow-up questions. By contrast, Data QnA provides self-service access to analytics without requiring technical knowledge.
With Data QnA, users of Google BigQuery, Cloud Storage, Bigtable, Cloud SQL, and Google Drive can ask free-form text questions like “What was the growth of product X last month?” and get answers interactively. They can also formulate analytical questions that Data QnA supplements with auto-suggested entities to return an English interpretation and an SQL query.
Data QnA enforces all underlying customer-defined data access policies, automatically restricting access to data to the right users. In addition, it offers a management interface for data owners or admins to define business terms for underlying data, which reports questions the users ask along with the answers and SQL query.
Data QnA is available at no additional cost for BigQuery customers in the U.S. and EU in English, with support for more regions and languages to follow. (All underlying queries and storage are charged as per the customer’s BigQuery costs.) Access in Sheets is through its Connected Sheets feature, which is included in G Suite Enterprise, G Suite Enterprise for Education, and G Suite Enterprise Essentials.
Companies can work with Google Cloud partners like Accenture, Deloitte, EPAM, Mavenwave, SADA, and Wipro to get started.
Author: Kyle Wiggers.
Source: Venturebeat