The first m-agricultural system in Ethiopia (agricultural interactive voice response system) Essay

We have defined and presented the problem definition along with its context by considering various perspectives. The rationale is also discussed based on the detail analysis of both claims and context of use. Here, we present the design solution and provide rigor through empirical evidence and practical evaluation.
One example solution for the interaction mechanism is voice interaction with minimum hierarchy navigation and keypad input. To check this solution empirically, we have chosen one project (agricultural mobile information system project in our context) recently deployed and functional throughout the entire country. It is the first m-agricultural system in Ethiopia (agricultural interactive voice response system) envisioned to empower smallholders living in the rural area with low-resource setting.
Currently this mobile application has two modules agronomic information enquire and notification message component (one is to pull information and the other is to push information). This automated hotline is free, providing information on a wide range of agricultural activities. Keypad menu options are used to select farmers' particular area of interest. This application also includes three local languages options (Amharic, Oromiffa, and Tigrigna).
We chose this project over a prototype since it is functional and easy to evaluate, enabling learning from experiences of use. We have conducted qualitative evaluation of the different functionalities to understand the solution space for the context.
Our preliminary evaluation result of the specified feature voice interaction mechanism showed promising result to produce proto-pattern [19] that can be further refined and produce complete pattern solution.
Usage statistics of the solution - in less than two years of the project period, there are about 1.27 million registered user and more than .92 million are identified themselves as farmers (who generally satisfy our user profile). These users made about 8.5 million information enquiry, which means on average a user used the system at least eight times. Considering the time and the content of the application, the number of user and usage seems very reasonable.
Observed challenges menu navigation (this application is eleven levels deep, out of which seven helps to build the profile (i.e., one to set the language, six to get the location of the user (country administrative levels that helps provide specific information to the location of the user), and the remaining four is for the actual content). Even within a menu itself there are a number of sub-menu choices (for instance one sub-menu has 18 options). The profile building process is only for the first time after that you can go directly to the content. In our evaluation, we observed that first-time profile building helps provide customized information but also took much time and keypad interaction to reach the content level that the user sometimes does not want.
We only keep qualitative data for the application but quantitative data (efficiency and effectiveness) will be measured for the new pattern emerged. In light of this, we have proposed a solution to avoid long navigational hierarchies (such as locating the user through technological means) to be tested in follow-up work.