Semantic Layer Based Reporting Integrated with NLP/NLG
Semantic Layer Based Reporting Integrated with NLP/NLG
System Architecture
Enterprise Semantic Layer
Relations between attributes, definition of metrics, mapping to database objects, user access rights to row/column data make up the core of semantic layer. When the focus is on analysis, definitions of metrics can get very complex. Dhiva’s semantic layer supports a range of powerful functions that also enables metrics to be parameterized. The Governor tool makes the job of setting all of this up fast and easy.
Powerful Dhiva-SQL Engine
Dhiva-SQL engine generates SQL that’s optimized and appropriate for the underlying database. The engine is aggregate aware, handles semi-additive and non-additive metrics, supports row level security, and allows parameterized metric evaluation. The resulting SQL query is optimized to reduce the number of passes and increase performance.
Integrated AI – NLP/NLG Analytical Layer
Dhiva Analytical layer enables the integration of NLP/NLG with the Semantic Layer. The KPI knowledge contained in the Semantic layer enables Dhiva Analyst to “understand” your data domains. Text generation (NLG) is enabled for any data structure without any template based customization - driven by dozens of cross-domain univariate and multi-variate analytical assertions.
Middleware Layer Using REST Services
All the functionality of platform is exposed as modular REST APIs enabling easy extendibility. The REST services include (but are not limited to) Report, Storyboard, Domain, Publish, Export, Analyst/NLG, Schedule, Security, Governor, Admin etc.
Intuitive Reporting & Visualization UX
With a responsive browser based UI built using Angular JS and HTML5, tool can be used seamlessly on mobile, tablet and desktop. The powerful export feature, enables the reports and dashboards to be exported in native office formats (not just images) or as an offline application where the look and feel of the application can be retained even while the snippets are shared with users with no access to the tool.
Cloud, On-Premise or Hybrid
Dhiva can be integrated within your data infrastructure with minimal effort and cost – and works with data sets of varying sizes (few MB to several PB) and designed in diverse forms (Star, Snowflake, 3NF, ROLAP etc.) It seamlessly works on cloud and on-premise with a number of databases, including Redshift, Snowflake, Teradata, Oracle, Hadoop, Snowflake, SQL Server, Postgres etc. With no additional infrastructure requirement, Dhiva platform can reduce TCO significantly.
More Reasons to use Dhiva-Platform
Intelligent
Dhiva can be configured for data metrics across domains. Analytical assertion and text generation engine integrated with NLG/NLP Smarter due to integration NLG engine with the Semantic Layer.
Quicker
Dhiva can be fit seamlessly into your enterprise. It is significantly faster to deploy when compared to traditional BI tools.
Cheaper
Dhiva is a flexible platform that fits into your existing data infrastructure. No new hardware or software is required.
Secure
Access to each user may be controlled by row or column at a fine grained level. Governor has complete control on what each user gets to see.
Supports Single Source of Truth
Core KPIs can be defined and managed by the Governor – providing enterprise level KPI governance capability – a critical need for maintaining a “single source of truth” enterprise BI platform.
Made for the Enterprise
It has all the enterprise features required to effectively promote, enable and control access to enterprise intelligence
Technical Requirements
Front End:
HTML 5.0 compatible browser: Chrome, IE, Edge, and Safari
Devices:
Desktop (PC or Mac), Tablets, and Android & iOS smartphones
Middleware:
IIS Server, PostgreSQL
Databases:
SQL Server, Oracle, PostgreSQL, Redshift, Teradata, Snowflake, etc.
- Front End: HTML 5.0 compatible browser: Chrome, IE, Edge, and Safari
- Devices: Desktop (PC or Mac), Tablets, and Android & iOS smartphones
- Databases: SQL Server, Oracle, PostgreSQL, Redshift, Teradata, Snowflake, etc.
