Zig Zag Technology: Narrative Intelligence

Inspiration

Motivated by our years of experience in national security, Zig Zag Technology is a commercial business with a social purpose to harness technology to protect society from harm. Founded by Dr. Stephen Anning, whose PhD research from the University of Southampton provides the theoretical foundation, we are pioneering "narrative intelligence" – a revolutionary approach that combines narrative methodologies with graph-based natural language processing to understand the complex ways humans communicate harm, violence, and extremism.

What it does

Narrative intelligence uniquely combines methodologies from the social sciences with a graph-based approach to natural language processing. Example methodologies include hostile narrative analysis, a novel approach from Steve Anning's PhD research, which aims to detect processes of violence legitimisation in natural language, and personal experience narrative analysis, which focuses on understanding a victim's experience of harm. These methodologies give compelling insights into experiences of violence and harm.

The graph-based approach to natural language processing responds to limitations of using statistical algorithms for these methodologies. These algorithms fail to account for the complexities of how humans communicate violence and harm. For example, one experiment found that market-leading algorithms fail to distinguish between the very different sentiments expressed in Hitler's Mein Kampf and Martin Luther King's "I Have a Dream" speech. Our graph-based approach brings these methodologies to life with compelling insights to much higher standards of rigour than the current state-of-the-art..

How we built it

Steve built the current prototype during his PhD and has continued to develop it through bootstrapping since then. We are now looking to take this prototype forward by further developing the narrative analysis methodologies and the graph-based approach to natural language processing.

Challenges we ran into

The main challenge is finding the time and resources to develop the product. The business is currently in a familiar bind: needing a product to build its business, and a business to build the product. The current build is based on bootstrapping, between business development and delivery against other priorities.

Another challenge has been communicating the idea and value of narrative intelligence. Feedback from early pitches was that the approach sounded too much like an academic project rather than an operationally valuable product. Much time, therefore, has been spent on reworking the pitch using business language without losing the underlying academic rigour, which is the foundation of Zig Zag's brand value.

Accomplishments that we're proud of

The main accomplishment is realising the idea and building the prototype while completing a PhD. Steve's background is not in computer science; he was an Infantry Officer in the British Army. He has built the product as a self-taught coder. The combination of his operational and technical expertise has provided invaluable insight into how to respond to the shortcomings of existing approaches to NLP.

What we learned

Too much for one page! The primary lesson is about learning how to navigate a sociotechnical value proposition. The value proposition in the filter bubble of Zig Zag is quite apparent and compelling. Nonetheless, we are too technical for operators; for technicians, our differentiation is not apparent, or too challenging to the status quo. We will need to give careful thought to our marketing messages to communicate our sociotechnical value to two distinctly different audiences.

What's next for Zig Zag Technology: Narrative Intelligence

We now seek investment and early adopters of narrative intelligence to take the business forward. We have customer champions in the Army, DSTL, a charity and a private security company. The prototype requires further development before being pilot-ready for them. We welcome the opportunity to continue developing our business with you.

Built With

• python
• rdf