With a somewhat long list of co-authors, the "Trellis paper" [3] (not to be confused with "Tetris" the game) is among the most exciting SLE papers I've come across thus far. Published this year and surprisingly spearheaded by a PhD student; Lars Hummelgren (and not some professor such as many past ACM SLE papers we've covered here on BA), this is such a terrific work touching on several important matters in contemporary Machine Learning research. Interestingly, Lars's second name almost feels like the HMMs he's researching here!

---[About Paper]:

The paper[3] is about a popular machine/statistical learning technique; HMMs; Hidden Markov Models, specifically about the sparse kind. It has a great mini intro to HMMs for newcomers too. As with most SLE papers, focus is on a new software language; Trellis in this case, which provably and empirically has proven to be more performant in time-series learning problems involving sparse datasets (arguably more representative of realistic problems). Trellis is a nonexecutable DSL[4] that takes a Trellis model of a sparse HMM and compiles it into a Python library with which CUDA-powered learning can be executed via a clean API[3]. Paper delves into the details of that process, as well as discussing interesting related previous work too.


---[REFS]:

1. https://www.kth.se/profile/larshum

2. https://scholar.google.com/citations?user=l6ppBDIAAAAJ&hl=en

3. Hummelgren, L., Palmkvist, V., Stjerna, L., Xu, X., Jaldén, J., & Broman, D. (2024, October). Trellis: A Domain-Specific Language for Hidden Markov Models with Sparse Transitions. In Proceedings of the 17th ACM SIGPLAN International Conference on Software Language Engineering (pp. 196-209). URL: https://dl.acm.org/doi/pdf/10.1145/3687997.3695641

4. Lutalo, Joseph Willrich. "Software Language Engineering-Text Processing Language Design, Implementation, Evaluation Methods." (2024). URL: https://www.preprints.org/frontend/manuscript/3903e4cd075074a7005cb705a5ef26c5/download_pub


#review #notes #acm #sle #jwl #phd

End of Year Greetings from our Internet President

We explore a novel approach to enhancing voice-operated personal assistants by integrating a lightweight text processing language, TEA, into our existing Question-Answer Knowledge Bases (QAKBs). This allows for dynamic, context-aware responses and multi-turn interactions, paving the way for smarter and more adaptable AI assistants. I invite you to read the abstract paper and share your thoughts! Also, looking forward to getting some support or a research grant to help further work we started years ago finally bringing together the VOSA and TEA language projects!

And thus we've adventured towards the new year of 2025.. through thick and thin, across terrain and hurdles of many kinds. Greetings to you that's been with us along this journey. Let's hope for and look forward to an even more productive, more impactful year ahead.

Alright, as we must indeed eventually earn our PhD and other doctorate honours, it is surely time to move on with our SLEing... 👇🏻👇🏻

It is early 2025 and we continue our review of ACM SLE papers in a streak that kicked off in 2024. Today I came across the first SLE paper specifically about Python, the programming language we all love and one of the most popular today. Paper is written by Aamir Farooq and a very credentialed SLE authority, [Assoc.] Prof. Vadim Zaytsev.

---[Brief Bio]:

Since 2021, Aamir Farooq is a Computer Science graduate student specialising in Software Engineering, at the Technical University of Denmark [1] though the paper cites his undergraduate affiliation at University of Twente in Netherlands [2]. Professionally he is a software designer and engineer [1].

Prof. Vadim Zaytsev is a true authority [3] in the SLE field, with numerous notable contributions across the field including having been an Editor in Chief of the SLEBoK[4] among many other things. He is an Associate Professor at the University of Twente in the Netherlands[5]. He is affiliated with the Faculty of Electrical Engineering, Mathematics, and Computer Science (EEMCS), specifically within the Computer Science department[6]. His research focuses on Software Language Engineering, Domain-Specific Languages, Grammarware, and Automation[7].

Prof. Zaytsev has a strong academic background with a PhD in Recovery, Convergence, and Documentation of Languages from the Vrije Universiteit in Amsterdam[8]. He has published extensively in the field and has an h-index of 63, indicating a significant impact in his research area[5].