Imagine you could read the mood of every customer who calls your support line, posts about your brand online, or comments on your TV show — in any language. That's what this project built: a system that detects emotions from text, speech, and video all at once, and makes sense of them together. Think of it as a universal mood detector that works across languages and channels, so companies can actually understand how their customers feel, not just what they say.
Companies today collect massive amounts of customer feedback — social media posts, call recordings, video reviews, chat logs — in multiple languages across multiple channels. But most analytics tools only scratch the surface: they count mentions and assign basic positive/negative scores to single-language text. The real question — how do customers actually feel, and why — remains buried in unstructured, multilingual, multi-format data that no single tool can make sense of.
An integrated Big Linked Data platform for emotion analysis across text, speech, video, and social media in multiple languages. Three commercial pilot applications were fully implemented: a Social TV system for analyzing viewer emotions, a Brand Reputation Management tool for tracking how people feel about brands across markets, and a Call Centre Operations system for detecting customer emotions during support interactions.
If you are a TV broadcaster dealing with fragmented viewer feedback across social media, live chat, and call-ins — this project developed a Social TV pilot that analyzes audience emotions in real time across multiple languages and channels. It lets you see not just what viewers are saying, but how they feel about specific segments or shows, helping you make programming decisions based on actual emotional engagement.
If you are a call center operator dealing with high volumes of multilingual customer interactions and struggling to spot dissatisfied callers before they churn — this project built a Call Centre Operations pilot that analyzes emotion from speech and text simultaneously. It processes calls across multiple languages, flagging emotional escalations and giving managers a real-time dashboard of customer sentiment.
If you are a brand manager dealing with reputation monitoring across markets in different languages — this project delivered a Brand Reputation Management pilot that fuses emotion signals from social media posts, comments, and video reviews into a single view. Instead of tracking mentions, you track how people actually feel about your brand across 5+ European markets simultaneously.
The project received EUR 3,036,910 in EU funding across 10 partners over 2 years. Licensing or deployment costs would depend on negotiation with the consortium partners. As an Innovation Action with commercial pilots already built, the technology is closer to market-ready pricing than a pure research prototype.
The platform was specifically designed for large-scale processing of heterogeneous big data streams — text, speech, video, and social media combined. The objective explicitly addresses scaling up to huge data volumes that existing commercial solutions cannot handle. Exact throughput benchmarks would need to be confirmed with the technical partners.
The consortium includes 4 SMEs and 4 universities across 5 countries (Ireland, Germany, Spain, Italy, Czech Republic). IP arrangements would have been defined in the consortium agreement. Contact the coordinator at University of Galway for licensing discussions.
Multilingual capability is a core design goal, not an add-on. The platform handles multilingual text, multilingual speech, and social media in different languages. The consortium spans 5 countries (CZ, DE, ES, IE, IT), suggesting at least those languages were tested.
Yes. Three commercial pilots were implemented with both initial and final versions: Social TV, Brand Reputation Management, and Call Centre Operations. These are documented across 6 demo deliverables, indicating iterative testing and refinement in real-world settings.
Most commercial sentiment tools analyze text in one language and output positive/negative scores. This platform fuses emotions from text, audio, video, and social network data simultaneously, works across multiple languages, and uses semantic knowledge graphs to connect emotional signals at the entity level — not just document level.
The platform was built as an integrated Big Linked Data solution with modular components. With 40 total deliverables including technical architecture and API documentation, integration pathways should be well-documented. The open-access data approach also supports interoperability with existing data pipelines.
The MixedEmotions consortium is well-balanced for commercialization with 10 partners across 5 countries (Ireland, Germany, Spain, Italy, Czech Republic). The 40% industry ratio — 4 industrial partners including 4 SMEs — signals genuine market pull, not just academic interest. University of Galway coordinated, bringing research depth, while the SME presence means smaller companies already vetted the technology for market fit. The geographic spread across major European markets (DE, ES, IT) is particularly relevant for a multilingual platform — the consortium literally speaks the languages the technology processes.
University of Galway, Ireland — reach out to their research office or use SciTransfer's matchmaking service for a warm introduction to the technical team.
Want to explore how multilingual emotion analytics could work for your customer data? SciTransfer can arrange a briefing with the MixedEmotions team and help you evaluate fit for your specific use case.
