Self-learning Service Recognition through Natural Language Processing
Understand Customer Requests automatically with Natural Language Processing
- In the recognition process a connection between the service to be recognised and the wording used in the customer request is presupposed.
- This is represented by trueAct NLP in the form of so-called patterns.
- A service to be recognised can consist of one or several patterns. Every pattern hereby consists of one or more words.
- If a customer request contains all words of a pattern, the pattern, and thus also the service, is recognised.
- If in one customer request several services are recognised, trueAct NLP delivers the most suitable service or the weighted list of potential services.
trueAct NLP learns automatically
Depending on the scope of services, the definition of suitable patterns can be quite challenging. Natural Language Processing of trueAct offers the possibility to define sample texts for every service. Based on these examples patterns are created automatically. Generally about 20 to 30 example texts per service should be used. New examples can be added anytime and from these, new patterns are generated, for example on the basis of incorrectly identified cases or due to new services.
Further Functions of trueAct Natural Language Processing:
- Automatic recognition and storage of words that are not suitable for the differentiation of services because they can stand for several services.
- Highlighting of words that should be excluded from pattern generation, such as for example surnames or addresses.
- Weighting of services according to their expected or effective occurrence.
- Manual completion and change of patterns .
- Evaluation of hit and error ratio per language and service.
High Degree of Automation in Customer Service thanks to Natural Language Processing
The aim of trueAct NLP is not primarily a hit rate of 100%. Actual hits should only support service processes beneficially. For this reason it is generally recommended to use the trueAct NLP method for most frequently used services. In this way a degree of automation is achieved that pays off quickly.
The following example shows the connection:
- If the 20 most important services can cover 80% of cases of incoming customer requests, the achieved hit rate per service amounts to 90%.
- The realisation of this cases results in an overall hit rate of 72% .
- The next smaller service will amount to hardly more than 2% of the cases .
- Its recognition would therefore improve the overall hit rate by less than 2 percentage points.
trueAct NLP for your Multichannel Management
trueAct NLP supports Multichannel Management in customer service optimally. Typical cases for application are for example:
- Email Response Management (including text messages/SMS, Fax and scanned mail)
- Multichannel Response Management (including chat and Social Media)
- Self-Service via user-friendly web applications
- Knowledge support for case processing (knowledge management)
trueAct NLP allows you to process your customer requests automatically and thus very quickly enables a high degree of automation and a fast return-on-investment.