1. Home
  2. Industries
  3. Manufacturing

How Siemens challenged Cortical.io Semantic Search
to leverage its Know How in Finance/Tax

Customer

Goal

Improve the information retrieval productivity across multiple knowledge databases

Challenges

  • The company operates multiple Intranet sites and expert knowledge bases for various business units and departments
  • These sites contain large amounts of diverse, partly unstructured documents that have been compiled over years and due to the varied nature of the documents, are difficult to search
  • The current search solutions do not yet support natural language queries
  • An internal study revealed that about 60% of inquiries to the Financial Reporting support center could have been answered with information available in the intranet

Solution

Leveraging Cortical.io patented approach to natural language understanding, the company implemented Semantic Search as a proof of concept for internal Siemens departments responsible for Financial Reporting and Tax & Customs. The solution

 

  • was trained with domain-specific language models in English for Financial Reporting and in German for Tax & Customs.

  • allows the domain experts to inspect and fine tune the search results through a dashboard that displays the semantic similarity between query and answer.

  • can intelligently process natural language queries while taking advantage of the specificity of keywords.

  • supports the end user with information that makes the AI-based search results fully explainable.

Results

  • For Custom & Tax, Cortical.io’s accuracy for natural language queries, compound terms and keywords increased the hit rate by 60%
  • For Financial Reporting, Cortical.io achieved a 22% higher accuracy compared to the existing non-semantic search engine and a 15% higher accuracy than the solution ranked second.

The proof of concept performed by Siemens showed that Cortical.io natural language processing technology with its domain-specific semantic language space was able to achieve significantly better search results in the area of taxes than keyword-based search engines. It has the potential to become a key technology for semantic use cases.

Darja Meyer, Head of Tax Tech Lab Germany, Siemens AG

Cortical.io demonstrated in our proof of concept that they have created an efficient search technology based on the Semantic Folding theory, and best fulfilled the expectations of our users for complex tax-specific search queries.

Florian Schelle, Head of Digital Finance Workplace, Siemens AG

See Contract Intelligence In Action - free demo