- Title
- An investigation into a natural language interface for contact centers
- Creator
- Sankar, Gopal Ravi
- Subject
- Call centers -- Management
- Subject
- Customer services -- Management
- Subject
- Telephone selling
- Date Issued
- 2009
- Date
- 2009
- Type
- Thesis
- Type
- Masters
- Type
- MSc
- Identifier
- vital:10468
- Identifier
- http://hdl.handle.net/10948/890
- Identifier
- Call centers -- Management
- Identifier
- Customer services -- Management
- Identifier
- Telephone selling
- Description
- Contact centres are the first point of contact between a company and a customer after the purchase of a product or service. These centres make use of contact centre agents to service customer queries. In the past contact centres hired as many agents as they could in order to service customers, which have led to an increase in personnel costs causing contact centres to become costly to run. Automation techniques were introduced to decrease personnel costs and one such technique is the Interactive Voice Response (IVR). The usability of IVR systems is, however, dismal. Customers would rather speak to a contact centre agent than navigate through the menu structure found in these systems. The menu structure has come under scrutiny because it is difficult to use and navigate, is often not aligned to caller usage patterns, and the menu options are long and vague. This research investigated whether a Natural Language Interface (NLI) could alleviate the problems inherent to IVR. NLIs, however, come with their own disadvantages of which the main ones are ambiguity and the loss of context of a conversation. Two prototypes were implemented, one of which resembled an IVR and the other an NLI (using ALICE concepts). An evaluation of two prototypes confirmed the advantages and disadvantages of these concepts in accordance to theory. A Hybrid prototype was proposed with the aid of two models. The model which proposed an NLI using a rule base was chosen for implementation. The Hybrid prototype was then evaluated against the NLI and IVR prototypes to deduce which prototype was the most effective, efficient and satisfying. The evaluation through the aid of descriptive and inferential statistics showed that the Hybrid prototype was the most usable prototype. The evaluation of the Hybrid prototype confirmed that a Hybrid approach could limit the shortcomings of IVR through the elimination of the menu structure found in these systems, thereby allowing users to state their queries in natural language. The incorporated rule base provided the Hybrid system with long term memory, eliminating one of the main disadvantages of NLIs.
- Format
- xii, 153 leaves
- Format
- Publisher
- Nelson Mandela Metropolitan University
- Publisher
- Faculty of Science
- Language
- English
- Rights
- Nelson Mandela Metropolitan University
- Hits: 840
- Visitors: 934
- Downloads: 103
Collections
NMMU School of Computer Science, Mathematics, Physics and Statistics
| Thumbnail | File | Description | Size | Format | |||
|---|---|---|---|---|---|---|---|
| View Details Download | SOURCEPDF | 4 MB | Adobe Acrobat PDF | View Details Download |