The future of consumption, powered by structured content
6 mins
RWS’s Tridion team, along with its partner Semantic Web Company (SWC), recently attended SEMANTiCS 2022 conference. It turned out to be a great opportunity for Tridion to showcase its solutions and also learn more about semantic AI from other participating companies.
We held our own session during this event, where Tridion’s Product Marketing Director Arpita Maity spoke about the growing importance of voice and chatbots powered by structured content at the backend. Our partner Semantic Web Company’s CEO and co-founder, Andreas Bluamer spoke at length about the critical role of knowledge-based recommendation systems in the digital workplace.
We had several other expert speakers from well-renowned global companies and key highlights from their respective talks have been covered in this blog.
It is interesting to note that each of the use cases and scenarios discussed during the conference highlights the importance and need for structured content.
Intelligent voice-assistant is the future of marketing and sales
Tridion’s Arpita Maity highlighted the way we humans are getting addicted to hands-free and off-screen ways of doing work. In 2020, 30% of web browsing sessions were screen-less, done via voice search. Voice is expected to be a $40 billion channel in 2023!
For smart speaker users “voice” leads them to the local shop or area to visit as 76% of them perform local search through such devices at least once per week. It is no surprise that 75% of voice search results are from the top 3 desktop results. This means that the content that ranks high on the desktop is used by voice assistant as the answer.
Voice search is thus closely tied up with desktop search and hence the same principles of structured content and SEO remain valid even for voice search. Given the paradigm shift towards voice search, it has become an absolutely vital channel to drive guided customer experiences.
RWS in general and Tridion provide the technology to deliver such experience with language translation and CCMS capabilities respectively. In particular structured content plays a vital role. Imagine content as Lego blocks built by authors, which run through reviewers only once. Such process repeats and several collections of content are created.
Now, when the user performs a query through voice, the assistant dives in and fetches the most relevant content quickly. In the background, semantic AI/knowledge graphs do the job of linking different content collections and serving it to the user as per his / her search context.
Arpita shared a simple real-life example, where a user named ‘Madalina’ faces an issue with her phone and uses voice assistance to solve the issue. Assistant arranges a pick from one of the nearest phone repair shops. The phone is then repaired and delivered back to Madalina.
In this example, several systems such as web customer portal, identity and access management, ERP and product inventory, knowledge base / structured content, sales and ordering product inventory, business process management, and language translation work in the backend.
B2B is moving in the same direction, Arpita shared:
Potential voice assisted applications in healthcare
And the common route for delivering such guided experiences to users passes through ‘structured content’:
Structured content enables guided experiences
Composite AI turns structured content into insights
Bernd Schopp, CEO and co-founder at Squirro, highlighted the fact that 80% of enterprise data is unstructured. While NLP and Machine Learning are heavily applied to analyze such data, it is not enough to provide insights for decision-making as it doesn’t incorporate domain knowledge.
Cognitive AI bridges this gap and delivers augmented intelligence that supports decision-making. Squirro works with Semantic Web Company’s Pool Party semantic suite to bring in the domain knowledge, which it terms as “Symbolic AI”. While, Squirro itself provides the NLP and ML, which it terms as “Data-driven AI”. And then “Composite AI” is the combination of these AI technologies, which ultimately delivers the insights that users want to make informed decisions.
Composite AI turns structured content into insights
Bernd Schopp, CEO and co-founder at Squirro, highlighted the fact that 80% of enterprise data is unstructured. While NLP and Machine Learning are heavily applied to analyze such data, it is not enough to provide insights for decision-making as it doesn’t incorporate domain knowledge.
Cognitive AI bridges this gap and delivers augmented intelligence that supports decision-making. Squirro works with Semantic Web Company’s Pool Party semantic suite to bring in the domain knowledge, which it terms as “Symbolic AI”. While, Squirro itself provides the NLP and ML, which it terms as “Data-driven AI”. And then “Composite AI” is the combination of these AI technologies, which ultimately delivers the insights that users want to make informed decisions.
Combining human expertise with machine learning to derive insights
Bernd shared the levels of augmentation and application at each level that can be reached using such cognitive AI. Beginning from intelligent search to insights to recommendations:
Levels of augmentation
Bernd also shared a couple of examples: 1. Library of Congress and 2. User wanting to cook bread. He demonstrated the process using what he terms as ‘Golden Triangle’. It takes into account the user’s ‘intent’, ‘context’ and the spectrum of ‘content’ available around it.
Intelligent search that provides solution
Zach Wahl, CEO at Enterprise Knowledge, LLC walked us through their experience in integrating machine learning, knowledge graphs, and enterprise search to deliver ‘next-level’ search experience. Enterprise Knowledge, LLC is a knowledge management consulting firm with more than 80 expert consultants.
Zach shared a very interesting and concise view of the way search has matured over time which demonstrates the evolution from basic keyword search to today’s contextual search:
Search maturity roadmap
He explained the various processes through which the content and data passes through to ultimately reach a form that enables user to make decisions. As expected, structured content and NLP – RWS and Tridion’s key expertise areas play a crucial role to furnish content in a form that analytics and aggregation can be performed seamlessly.
Enterprise content processing cycle to support decision making
Future of digital workspace – knowledge-based recommendation systems
Andreas Bluamer (Semantic Web Company) drew our attention towards the limitation of ecommerce platforms, which tend to recommend more of the same, while they should be recommending the next stage of purchase. He took an example of a user purchasing couch, ideally the recommendation engine should recommend some other furniture that is used in the living room, may be a coffee table.
Andreas shares a number of use cases where knowledge recommender works to deliver benefits across different industry sectors. Some of the examples he shared include:
- For a management consulting company it built a HR recommender (a semantic matchmaking tool based on knowledge graph) to connect employees with their co-workers and show then relevant projects
- For a healthcare company SWC built a recommender engine to improve effectiveness of its learning platform by delivering most relevant courses to the users and driving engagement
Andreas highlighted the evolution that organization needs to go through to build such recommender systems and without doubt structured content is ‘the founding stone’ to get the journey started.
Using knowledge graphs and industry standards to help interlink modern industries
Atanas Kiryakov, CEO at Ontotext shared some examples in which knowledge graphs can be used to link industries. E.g. he mentioned that knowledge graphs can used to link: electricity (smart grid), water (smart water), building automation systems, smart cities, and logistics (maintenance / product delivery). He went on to share some of the use cases and example companies across these sectors.
Knowledge graphs to bridge gap between data and design
Adam Keresztes, business owner at IKEA Knowledge Graph shared an interesting view about UX designers’ limited knowledge about the underlying data / content. He believes that knowledge graphs have the potential to make data discoverable and understandable to UX product designers in a humanistic way. And when done right it will help UX designers deliver better products and enhance customer experience.
Knowledge graphs market has matured
Atanas Kiryakov CEO of Ontotext held a panel discussion with four CEOs of leading software vendors in the market. All of them unanimously agreed that the market has moved beyond ‘if’ and ‘why’ knowledge graphs to ‘how’ and ‘when’. The panel reflected on the state of play, jointly analyzed typical implementation paths, discussed about the Dos and Don’ts and anticipated emerging trends.
All in all it was a highly enriching experience to learn about the most promising technology in the world of content from experts around the world with diverse background.
If you'd like to learn more about semantic AI, click here.
Wali Naderi
All from Wali NaderiRelated Articles
