NLP: Unlock the Hidden Business Value in Voice Communications
See where NLP fits into analytics.
Join the DZone community and get the full member experience.Join For Free
Today organizations capture an enormous amount of information in spoken conversations, from routine customer service calls to sophisticated claims processing interactions in finance and healthcare. But most of this information remains hidden and unused due to the difficulty of turning these conversations into meaningful data that can be effectively analyzed through Natural Language Processing (NLP).
Simply applying speech recognition software to voice conversations often results in unreliable data. State-of-the-art speech recognition systems still have trouble distinguishing between homophones (words with the same pronunciation, but different meanings), as well as the difference between proper names (i.e. people, products) and separate words. In addition, there is also the challenge of identifying domain-specific words accurately. Thus, in most cases, using speech recognition software alone doesn’t produce accurate enough data for reliable NLP.
Domain-specific taxonomies are key to understanding conversations via speech recognition systems. With them, we can feed conversations to knowledge graphs that understand the conversation and make connections in the data. Knowledge graphs provide the ability to extract the correct meaning of text from conversations and connect concepts in order to add business value.
Knowledge graphs fed with NLP provide two prime opportunities for monetization. First, organizations can better understand their customers to improve products and services more to their liking, which in turn boosts marketing, sales and customer retention rates. Secondly, this analysis gives contact center agents real-time support for optimizing customer interactions to produce faster resolutions, better conversion rates, and cross-selling and up-selling opportunities. These approaches enable companies to capitalize on speech recognition knowledge graphs, accelerate their ROI, and expand their bottom lines.
Taxonomy Driven Speech Recognition
The story of taxonomy-driven speech recognition closely relates to knowledge graphs. The first wave of knowledge graphs was built from taking structured data and turning it into semantic graphs that support the linked open data movement. The next wave is all about unstructured data. People started doing Natural Language Processing on documents and textual conversations like emails and chats. Doing so accurately for a given domain requires a taxonomy to understand the words and concepts. Otherwise, downstream processes like entity extraction and event detection won’t work.
The next step was analyzing all the dark data in voice conversations. But the problem is that current speech recognizers have difficulty with these interactions because there are too many domain-specific words. Supplementing these speech recognizers with domain-specific taxonomies drastically improve their output because they know what words to expect. For example, if there was a conversation about medicine or company products, conventional speech recognizers are likely to confuse domain-specific words. But by training speech recognizers with taxonomies, the output is much better, as is the subsequent NLP for storing more accurate information in knowledge graphs.
Once these conversations are in intelligent knowledge graphs that understand them, organizations can begin profiting from this information. Understanding customers in their own terms about what they do or don’t like, the products they mention, and the competitors they reference improves analytics for long and short-term strategies. These analytics use cases vary according to need. Quality control—ensuring agents talked about the right things with appropriate responses to conversations—is particularly useful.
Analytics are also helpful for assessing metrics about specific clients. Contact centers can analyze conversations with customers to see how many times certain products were mentioned, those of competitors were mentioned, and how this information relates to closing or missing out on sales, for example. These findings are redeemable for everything from sales and marketing strategies to product research and development.
Real-Time Agent Support
Analyzing customer interactions with speech recognition knowledge graphs substantially boosts the performance of contact center agents by giving them real-time access to timely information. For example, if a customer talks about a competitor, systems can bring up a “battle” card. Or, if a customer happens to mention a product, based on that term—and the underlying taxonomy supporting understanding of it—agents can bring up a screen with all the information about that product and related technologies.
These types of analytics are perfect for issuing recommendations based on rules and machine learning, which increase successful up-selling and cross-selling opportunities. This same approach is useful for resolving disputes or troubleshooting products and services, which decreases the likelihood of churn while fostering customer loyalty.
The Business Value of Taxonomy Driven Speech Recognition
The merit of these business value applications of speech recognition and text analytics stems from the taxonomy foundation that is core to knowledge graphs. Clarifying the relevant words, their meaning, synonyms, and hierarchies of meaning for a specific domain deliver maximum organizational utility for speech recognition capabilities. Such taxonomies deliver the best results for speech recognizers deriving text from spoken words, and are essential for extracting entities pertaining to marketing, sales, call types, competitor references, and everything else.
By coupling these capabilities in knowledge graphs, organizations get the additional benefit of readily traversing them for everything relevant to specific use cases— from issuing recommendations to collateral material to competitor shortfalls, and even successful previous agent approaches with customer segments. These repositories are the most effective means of democratizing this knowledge and applying it outside of speech recognition to any text analytics.
Opinions expressed by DZone contributors are their own.