Unifying FSM-inference algorithms through declarative specification

Abstract

Abstract-Logging system behavior is a staple development practice. Numerous powerful model inference algorithms have been proposed to aid developers in log analysis and system understanding. Unfortunately, existing algorithms are difficult to understand, extend, and compare. This paper presents InvariMint, an approach to specify model inference algorithms declaratively. We apply InvariMint to two model inference algorithms and present evaluation results to illustrate that InvariMint (1) leads to new fundamental insights and better understanding of existing algorithms, (2) simplifies creation of new algorithms, including hybrids that extend existing algorithms, and (3) makes it easy to compare and contrast previously published algorithms. Finally, InvariMint's declarative approach can outperform equivalent procedural algorithms.

Automatic mining of specifications from invocation traces and method invariants

“…Synoptic [4] uses the CEGAR [8] approach to create a coarse initial model, and then refine it using counterexamples that falsify temporal invariants. Meanwhile InvariMint [3] presents a declarative specification language for expressing model-inference algorithms, and improves the efficiency of algorithms, but neither their precision nor recall.…”

“…Such a specification should, for example, assert that communication libraries (e.g., Apache's [1] libraries) require a connection to be established before data is sent. Numerous techniques have been proposed to that end [3,4,10,15,19,20,32,33,41,46]. These techniques either (1) infer finite state machine (FSM) models that match the observed executions [3,4,9,20,32,38,44], or (2) identify the declarative class and method invariants by considering a library's state (i.e., its internal variables) [10,19,45].…”

“…To explore these strategies and evaluate the quality of the resulting models, we have developed two novel techniques: state-enhanced k-tail (SEKT) and trace-enhanced MTS inference (TEMI). SEKT improves upon an existing algorithm [5], which has been employed by many model-inference techniques (e.g., [3,32,38]), while TEMI infers state invariants [19] and uses them to produce a modal transition system (MTS) [27] of a library's inferred and observed executions.…”

“…Thus, our results are generalizable to little-used and poorly-known libraries. (3) In addition to modeling the behavior that avoids runtime errors, our techniques also model "good-practice" behavior. Good practice reflects a li-brary's intended use, although failure to follow it need not always result in runtime errors.…”

“…An empirical evaluation that indicates improvements in precision and recall over existing techniques. 3. Results that generalize to poorly-understood libraries.…”