.png)
1. Understanding the Experts' Thought Process
Having settled the retail sector and the most important metrics to consider for the expert system, this week's interviews had a bigger focus on the thought process used by the experts to solve the problem, including metric analysis and possible action plans.
First Interview
On the first interview, both the team and the experts agreed that the best way to capture and understand the thought process behind the metric analysis and customer satisfaction diagnosis should be through an approximation to the business rules that will compose the expert system.
As such, the experts started by indicating the intervals of expected values for each metric, which will later help the expert system determine if a certain metric's value is positive or not. Furthermore, having agreed with the team that certainty factors make sense in this problem, the experts also suggested the certainty factors associated to each metric, in case its value has either a positive or negative impact in the conclusion (i.e. satisfaction degree).
This way, the team can then write business rules for the expert system that may follow a similar format to the following:
If the value of metric X is not an expected value
Then the client is unsatisfied with a certainty factor of Y%
Second Interview
On a second interview, the team started by reviewing the values provided by the experts and suggested the definition of more complex rules (e.g. through the aggregation of metrics), that would result in fewer but better rules that could also be more easily managed and extended. The experts promptly agreed with this approach, although not much could be defined this week, but will certainly be a priority in the following interviews.
On the other hand, being the final phase of the problem, the team finished the interview by discussing some possible action plans with the experts, with both experts suggesting that the expert system takes into account two types of action plans: (i) plans that are applied to a single client (e.g. 10% discount voucher), (ii) plans that have a broader scope and take into account the frequency of the dissatisfaction causes (e.g. many clients are getting their payment methods declined, the company should do something about it). On the following interviews, more concrete action plans will be settled and integrated in the knowledge base.
2. Knowledge Engineering's Report and Presentation
Being the fourth week of the project, the teachers asked for a presentation and technical report that described the project's current progress, namely the obtained knowledge and the strategy used by the team to acquire it.
Having both the interviews and the report/presentation to prepare, the team had to be very organised in order to have time for everything, but in the end, the presentation seemed to catch people's interest and the report ended up with a very detailed representation of the current knowledge base, although it is not finished and will be expanded for the final delivery later this month.
3. Week Retrospective
In retrospective, the past week went rather well, just like the previous one, with the team having achieved every planned goal and, hopefully, catching people's interest in our project based on our presentation.
Next week, having obtained a solid knowledge base and finished this week's presentation and report, the team must start developing the expert system and, afterwards, the web application that will allow the demonstration of a practical case in the final presentation and Demo Day's pitch.
Comments