There’s only one affordable way to provide customer support in many languages: deploy a Multilingual Knowledge System (MKS) to access your, primarily monolingual, knowledge base and use the MKS as an information infrastructure for the process and its tools.

The requirements list for global customer support is long. Actually, a product for real global customer support still has to be built. The solution architecture below illustrates how. The chart shows the data flow, the different building blocks, and how the MKS makes them work multilingual.

Listen and Speak to your Customer

In the support section of your web site customers describe their problems, often in their mother tongue. Particularly younger customers might simply write about their issues on social media. Mobile customers want to use either voice or chat to communicate with support. No matter how the customers contacts you, they expect to be answered via the same channel in the same language.

Cross-language Semantic Search

The textual data (in case of voice gained through automatic speech recognition) must be tokenized for the following analytics into words, numbers, special characters, etc. Named entities, measurements, product names, corporate terminology, etc. need to be marked. When retrieved from social media text analytics decides how the information has to be further processed. Ideally the author and customer can be identified in the structured customer data. The Advanced Linguistic Search (ALS) finds then the relevant information in the Knowledge Base. ALS combined with a Knowledge System searches semantically; combined with a MKS it can also search cross-language.

One Corporate Language, many Customer Languages

The Multilingual Knowledge System enables companies to retrieve knowledge from many languages. On top of Advanced Linguistic Search and MKS an analytic component can answer questions. Only ontology/taxonomy combined with advanced search provides the intelligent responses that customers and agents need. Machine Translation is used when the solution, a question, or dialogue is passed back to the customer in her language.

Support the Support Agent

In many standard case a Question & Answering System can detect that crucial information is missing. It then formulates the respective questions. A dialogue system is customized with domain and company specific templated dialogues. It can talk with the customer.Today’s maturity level of speech technologies, natural language processing, artificial intelligence, dialogue systems, machine translation, and text analytics cannot handle all situations without human intervention. Luckily it knows when it needs help. It then simply handled the case over to a human agent. At that moment most standard stuff is already dealt with. The agent has all relevant information at hand and can focus on solving the customer’s issue.

MKS, Information Infrastructure for Customer Support

The Multilingual Knowledge System is the central information infrastructure for the global customer support solution. The product knowledge stored in the MKS is used to customize speech recognition, voice generation, NLP, text analysis, and dialogue systems. It tunes the Machine Translation. It provides the multilingual terms and knowledge for the semantic cross-language search in the Knowledge Base.
In order to enhance the company’s Knowledge Base, multilingual Text Analytics combined with the intelligence of the MKS provides quantified information about product and process defects reported in different languages on any media. This helps Product Management to prioritize features or even come up with new ones. It supports Marketing in positioning new product capabilities and understand customer use cases. For all customers, globally!

By Jochen Hummel CEO of Coreon GmbH and ESTeam AB


Jochen is co-founder and CEO of Coreon GmbH. He also heads ESTeam AB, a leading LangOps company. As founder of TRADOS he had established Computer-Assisted Translation as a product category, with Semiox he is reinventing multilingual again. He is an internationally known software executive, business angel, and serial entrepreneur.