Customer Intelligence

Our emphasis was on capturing the opinion of customers, esp. consumers, wrt a given brand or product within a given industry. We chose as the context retail, which is in the midst of moving from physical stores to digital stores or perhaps their combination. The angles we chose were

Text Analytics for Customer Intelligence

Text Analytics

Customer Intelligence Context

Enables automated monitoring of customer opinion and its changes over time.

Problem

To measure the attitude and its change over time of consumers within some particular industry.

Solution

Benefits

Ability to measure the success or failure of communication and marketing effort wrt a brand or product. Early detection of shifts in consumer opinion which might indicate quality issues or active effort by competitors.

Context Intelligence for Marketing

Context Intelligence for Marketing

Customer Intelligence Context

Enables single view intuitive analysis of consumer opinion in a given industry.

Problem

Perception of the global context for business critical events wrt consumer opinion in the industry is very difficult for a global company.

Solution

Text analysis combined with intuitive visualizations for the recognition of business-relevant events:

Benefits

Ability to detect early and react to threats and opportunities in the market for operative management, timely competitive intelligence about consumer opinion in B2C business, and ability to discover new ideas globally for R&D.

BI-Relevant Event Detection in Media Data

BI-Relevant Event Detection

Customer Intelligence Context

Enables automated cross-lingual and cross-industry detection of probably relevant emergent customer opinion trends.

Problem

Catching emergent trends in an industry, e.g. new consumer needs, are the more valuable the earlier they are detected.

Solution

Real-time statistical analysis of streaming documents for detecting deviations in content over time. This provides the ability to generate alerts based on detected deviations.

Benefits

Ability to react to non-evident threats and opportunities ahead of competition.

Transportation Behavior Monitoring

Transportation Behavior Monitoring

Customer Intelligence Context

Enables automated detection of the mode of customer movement while moving between home, work and retail stores to capture customer behavior wrt mode of transportation.

Problem

A smartphone based system is needed for detecting the current transportation mode of the user from accelerometer measurements. Developed techniques can be used to construct acceleration profiles for different vehicles, useful, e.g., in assessing vehicle types, skill level of drivers, or even fuel consumption of vehicles.

Solution

Analyzing accelerometer signals to extract movement profiles that capture characteristics acceleration and breaking patterns of different transportation modalities. By matching observed patterns against these signatures, the system can then determine the transportation modality of the user.

Results

> 90% accuracy for separating between pedestrian and vehicular motion, and > 80% accuracy for distinguishing different vehicles / public transportation.

Benefits

Context-Aware Predictive Model of Bus Ridership

Bus Ridership

Customer Intelligence Context

Enables prediction of customer movement between home, work and retail stores given the context wrt. bus ridership as an example of more detailed analysis.

Problem

Predict bus ridership as a function of time, weather, and other contextual attributes.

Solution

Predictions made using a Gaussian process regression framework.

Benefits

The resulting accurate predictive models are essential, e.g., for developing on-demand transit services.

Autopropagation Model for Indoor WiFi Locationing

Indoor WiFi Locationing

Customer Intelligence Context

Enables support for behavioral analysis of customers in retail stores.

Problem

Solution

Benefits

Intelligent Retail

Intelligent Retail

Customer Intelligence Context

Enables proactive analysis of customer behavior once the customer has reached a retail store.

Problem

Pathway analytics i.e. behavioral profiling and prediction of movements of people within retail environments.

Solution

Benefits

E.g. ability to promote sales based on customer preferences given product placement.

Gesture Interaction

Gesture Interaction

Customer Intelligence Context

Enables powerful combination of physical store and digital content.

Problem

Solution and Results

Gestimator: segment-based gesture recognition for complex gestures (> 95% accuracy).

Benefits

Enabling retail store of the future.

Screenshots and Promotional Videos

Selected Publications

S. Bhattacharya, S. Phithakkitnukoon, P. Nurmi, A. Klami, M. Veloso & C. Bento (2013). Gaussian Process-based Predictive Modeling for Bus Ridership. Proceedings of the 3rd International Workshop on Pervasive Urban Applications (PURBA).

S. Hemminki, P. Nurmi & S. Tarkoma (2013). Accelerometer-Based Transportation Mode Detection on Smartphones. Proceedings of the 11th ACM Conference on Embedded Networked Sensor Systems (SenSys).

T. Pulkkinen, J. Verwijnen & P. Nurmi (2015). WiFi Positioning with Propagation-based Calibration. roceedings of the 14th International Symposium on Information Processing in Sensor Networks, April 13-16, 2015 Seattle, WA, USA. doi:10.1145/2737095.2737144

T. Pulkkinen & J. Verwijnen (2015). Evaluating Indoor Positioning Errors. Proceedings of the International Conference of ICT Convergence (ICTC) 2015, Oct 25-30, Jeju Island, Korea.