Artificial Intelligence in Business Information:

The Opportunity & Challenge of Structuring Knowledge

 

Written by Nicolas Bombourg, Managing Director and Co-founder of Findout Ltd

 

Developing Artificial Intelligence (AI) requires continuous research and testing. As the Managing Director of Findout Ltd, we have made this process a standard in developing advanced AI technology for business users seeking information. Our technology structures a vast amount of raw material into knowledge that is more valuable thus helping business owners make decisions faster.

Artificial Intelligence, in the business information industry, is all about structuring knowledge. And the raw material is data.

Where does the data volume explosion comes from?

The raw material is a vast abyss of published information and the volume is estimated to see an 800%+ increase in the next five years according to Gartner. Information is constantly being posted on the web, tweeted, and ‘liked’.

Gartner also estimated that 80% of the data volume explosion will be unstructured data from sources like PDF, PowerPoint, Word, videos, social media (i.e. LinkedIn, Facebook, Google+, Twitter, YouTube) and others. The remaining 20% of data is already structured in Excel and CSV formats or any other file that can be easily manipulated or transformed into graphics.

Working with unstructured data presents challenges. For example, videos contain valuable business information but they are in image and voice formats. For non-machine-readable content, such as PDF, information cannot be easily extracted to produce valuable information-n. These forms of publicly available unstructured data include legal documents, annual reports, or investor presentations.

Why is this an opportunity?

A “normal” human could not easily analyse such data manually. Therefore, there is a critical need for data technologies to transform this formless data into opportunities and strategic assets. Spotify, for example, has taken this opportunity to transform data to create an ideal playlist using AI over a Big Data stream. The more information that feeds the stream, the greater the ability to make a better decision.

Up until now, the technologies were individually operated blocks perceiving the environment and giving it meaning. But today, we are able to make them work together, to solve a small class of vertical problems. The problems that are typically solved are those that are already mastered such as driving a car.

Most of the companies that claim to work on the Artificial Intelligence field are at this stage actually:

  • Predictive models = news recommendations for example.
  • Automatic data discovery = pattern & concept identifications, new trends or new indicator in a text corpus.
  • Natural language generation: machine translation, or question answering.
  • Perception / recognition: image & speech recognition for example.

How does Artificial Intelligence work?

Because AI works like human intelligence, the same methods are used to structure the knowledge.

As a child, humans learn by collecting information from what is seen and heard to, then, create assumptions. These assumptions are retrieved and tested to determine if it is true…or not. The adults, then, correct misunderstandings. Once corrected, knowledge is assimilated and structured.

Similarly, behind the scenes, a machine takes unstructured information through pattern detection algorithms to generate the assumptions, which are then checked and validated as structured knowledge. New assumptions are continuously inserted into the system to increase precision. The difference between human and machine learning is the volume of data that can be understood and analysed as insights.

How does Artificial Intelligence help Information Professionals?

Using AI systems for business information falls into two broad categories. There are those that embed AI to ameliorate the end-customer experience (e.g. publishers embed recommendation engines into their web properties to drive paid subscriptions). Then there are those that incorporate AI into a workflow to uncover insights for informed decisions (e.g. Novartis following the development projects of thousands of pharmaceutical companies, in the real time).

In conclusion, AI can now manage the massive amounts of unstructured knowledge, disambiguate terms and master human tasks. Though these systems are highly advanced, human analysts like data scientists, engineers and UX designers, remain the cornerstone of the system. Structuring knowledge is still a human job.

 

 

Spread the word. Share this post!

Site last updated July 7, 2017 @ 8:47 am; This content last updated January 11, 2017 @ 4:10 pm