Experiences on Teaching Alloy with an Automated Assessment Platform

Abstract

This paper presents Alloy4Fun, a web application that enables online editing and sharing of Alloy models and instances (including dynamic ones developed with the Electrum extension), to be used mainly in an educational context. By introducing secret paragraphs and commands in the models, Alloy4Fun allows the distribution and automated assessment of simple specification challenges, a mechanism that enables students to learn the language at their own pace. Alloy4Fun stores all versions of shared and analyzed models, as well as derivation trees that depict how they evolved over time: this wealth of information can be mined by researchers or tutors to identify, for example, learning breakdowns in the class or typical mistakes made by Alloy users. Alloy4Fun has been used in formal methods graduate courses for two years and for the latest edition we present results regarding its adoption by the students, as well as preliminary insights regarding the most common bottlenecks when learning Alloy (and Electrum).

Bounded Exhaustive Search of Alloy Specification Repairs

“…We now assess our technique for automated repair of Alloy specifications. Our evaluation is based on two benchmarks of real faulty Alloy specifications, one taken from [3] and used in the evaluation of ARepair [19], and the other originated in the Alloy4Fun project [21], which includes 6 new models, with a total of 1936 faulty variants (considering different specification assignments resolved by different students). All the presented experiments were run on a 3.6GHz Intel Core i7 processor with 16 GB RAM, running GNU/Linux.…”

“…We evaluate our technique on a benchmark of Alloy specifications, including specifications previously used in assessing ARepair [19], [20], and a large benchmark of faulty Alloy specifications produced by students [21]. Our evaluation shows that our pruning technique significantly reduces specification repair running times, duplicating the number of repairs that can be produced within a 1-hour timeout, and reducing the repair time by 62X, on average.…”