ICSE Workshop on Software Engineering for Adaptive and Self-Managing Systems

About ICSE Workshop on Software Engineering for Adaptive and Self-Managing Systems

Overview — the SEAMS workshop and symposium proceedings

The ICSE Workshop on Software Engineering for Adaptive and Self-Managing Systems is the proceedings record for SEAMS, the specialist meeting on software engineering for systems that monitor themselves, reason about change, and adapt their own behaviour at run time. It is a conference proceedings series, not a journal. The JournalsWorld record lists the publisher as IEEE Computer Society, the United States, the subject area Computer Science, and two ISSNs, 2156-7891 and 2157-2305, together with a stored H-index of 25 and an SJR value of 0.371. In practice SEAMS papers have appeared through both IEEE and ACM channels over its history, reflecting its origin as a workshop of the International Conference on Software Engineering (ICSE) and its later life as an international symposium co-located with major software engineering conferences. The record title preserves the early ICSE workshop name; the community itself now calls the event the International Symposium on Software Engineering for Adaptive and Self-Managing Systems, while keeping the SEAMS short name that authors, readers and citation databases recognise.

SEAMS was established in 2006 at ICSE in Shanghai. It brought together several earlier workshop strands that had been meeting at ICSE and at the Foundations of Software Engineering conference since 2002, including workshops on self-healing and self-managed systems, on design and evolution of autonomic application software, and on architecting dependable systems. That origin matters. Before 2006, researchers interested in self-adaptation were scattered across dependability, architecture, autonomic computing and software engineering venues. SEAMS gave them a single recurring forum with a shared vocabulary, a shared set of example systems, and a programme committee able to judge both the software engineering contribution and the adaptation logic of a submission. Almost two decades later, that consolidation is the event's main historical achievement.

From workshop to symposium

The early editions followed ICSE around the world: Shanghai in 2006, Minneapolis in 2007, Leipzig in 2008, Vancouver in 2009 and Cape Town in 2010. The community then continued through Waikiki in 2011, Zurich in 2012 and San Francisco in 2013, by which point the meeting was being described as a symposium rather than a workshop — a change of label that reflected its size, its standalone programme and its archival proceedings, not a change of subject. Hyderabad hosted the ninth edition in 2014, Florence the tenth in 2015, Austin the eleventh in 2016, Buenos Aires the twelfth in 2017 and Gothenburg the thirteenth in 2018, in association with ICSE 2018. Later editions have continued the same pattern of co-location with the international software engineering calendar and, in the pandemic years, of virtual and hybrid participation. Edition numbering, host cities and co-location are announced on each year's official conference site, which is the authoritative source for a given year's dates and deadlines; this profile deliberately does not freeze a future edition's details into a series-level record.

Growth from workshop to symposium also changed what a SEAMS paper can be. Early workshops leaned toward position papers, architectural proposals and early experience reports from a field still defining itself. The symposium programme came to include full research papers, short papers, artifact contributions and doctoral or early-stage work, together with keynotes and, in some years, a most influential paper reflection looking back at work from a decade earlier. The 2020 programme, for example, recognised most influential paper awards for SEAMS work from 2006 and 2007, on architecture-based self-adaptation in the presence of multiple objectives and on an architectural style for solving computationally intensive problems on large networks. That retrospective habit is characteristic of the community: exemplars and older approaches are revisited, reimplemented and compared, rather than simply displaced by the next technique.

The research problem SEAMS addresses

A self-adaptive system changes its structure, parameters or behaviour while it is running, in response to changes in its environment, its goals, its resources or its own internal state. A self-managing system carries more of the burden of configuration, healing, optimisation and protection that would otherwise fall on a human operator. The engineering difficulty is that adaptation itself must be designed, analysed, tested and assured. An adaptation that improves response time may damage security; a healing action that restores one component may overload another; a learning component may drift away from the behaviour its operators certified. SEAMS exists because those are software engineering problems — problems of requirements, architecture, modelling, verification, testing, deployment and evolution — and not only problems of control theory or machine learning in isolation.

The classic organising device in the field is the feedback loop, most often described through monitoring, analysis, planning and execution over shared knowledge. SEAMS papers use that loop as a starting point and then interrogate it: What should be monitored, and at what cost? How is uncertainty represented when sensor data are incomplete or goals conflict? When should adaptation be centralised, and when should it be decentralised across many components, each adapting with partial information? How can an adaptation decision be explained to an operator, logged for later audit, or constrained so that a system never adapts itself into an unsafe state? How should engineers test a system whose behaviour is deliberately not fixed at design time? Those questions, rather than any single algorithm, define the proceedings.

Scope and recurring themes

Foundations and modelling. Formal and semi-formal models of adaptation, feedback-loop architectures, models at run time, goal models, and languages for specifying adaptation policies and guarantees.

Requirements and decision making. Eliciting and representing goals that can change at run time, handling conflicting quality attributes such as performance, energy, cost, reliability and security, and making trade-off decisions under uncertainty.

Architecture. Architecture-based self-adaptation remains the historical core of SEAMS. Papers examine architectural styles, connectors, component frameworks, decentralised adaptation, and the relationship between a system's architecture description and the running system it governs.

Verification, testing and assurance. Proving or testing properties of adaptive behaviour, runtime verification, simulation of adaptation, safety and security cases for systems that change themselves, and methods for giving operators justified confidence in an adaptation mechanism.

Self-healing, self-optimisation and self-protection. The original autonomic properties continue as application themes: detecting and repairing faults, tuning performance and resource use, configuring new components, and responding to attacks or failures without waiting for manual intervention.

Domains and exemplars. Cloud and service systems, cyber-physical systems, robotics, mobile and ubiquitous computing, web services and industrial systems recur as evaluation settings. Community exemplars — shared example systems that many groups implement — allow results from different teams to be compared on a common problem, and SEAMS has been an important home for that practice.

Human and societal questions. More recent programmes include work on human involvement in adaptation, ethics of autonomous and self-adaptive systems, and the point at which a human operator should be able to intervene, override or take responsibility. A 2020 panel on ethics for autonomous and self-adaptive systems illustrates how far the agenda has widened from its original architecture workshops.

What the proceedings contain, year by year

The 2007 proceedings show the field in formation: an opening paper by the organising group — Betty H. C. Cheng, Rogério de Lemos, Stephen Fickas, David Garlan, Marin Litoiu, Jeff Magee, Hausi A. Müller and Richard Taylor among them — followed by work on architectural styles for adaptive systems, autonomic element architectures, and mobility in software architectures. The 2009 Vancouver proceedings, issued with IEEE cataloguing and ISBN 978-1-4244-3724-5, organised papers around self-adaptive services, implementing self-adaptive systems and self-organising systems, including service management models, service-level agreement protection in virtualised data centres, self-healing composition of web services, and frameworks for developing self-managing Java-based systems. The 2010 Cape Town proceedings, published through ACM, included influential work on decentralised self-adaptation that is still cited as a orientation point for the trade-offs involved. By the 2013 San Francisco symposium, the programme included keynotes on science in the cloud and on a ten-year perspective on software engineering for self-adaptive systems, alongside papers on feedback loops, architecture-based adaptation in industrial systems, quality-of-service modelling for the cloud, formal correctness of dynamic updates, genetic approaches to mobile application adaptation, and diagnosing architectural failures at run time. That breadth — foundations, cloud services, mobile systems and formal methods in one volume — is typical rather than exceptional.

The 2014 Hyderabad proceedings were edited by Gregor Engels and Nelly Bencomo and published by ACM with ISBN 978-1-4503-2864-7. The 2019 Montreal edition, as the fourteenth international symposium, presented a full technical programme with learning, planning and traffic and robotics applications among its sessions, together with artifact evaluation — a sign of the wider software engineering community's turn toward reproducible computational results. Artifact papers at SEAMS 2018 were also archived through the Dagstuhl Artifacts Series, giving evaluated software and data a citable home outside the proceedings PDF. Authors considering a submission should therefore think about artifacts from the start: an adaptation engine, an exemplar implementation or a dataset that others can run is often part of the contribution, not an afterthought.

How to publish in SEAMS

SEAMS is a competitive peer-reviewed conference venue. Authors submit a paper for a specific edition by that edition's deadline, in the format and page limit stated in that year's call for papers, through the submission system named in the call. Papers are reviewed by the programme committee for the significance of the software engineering contribution, the soundness of the adaptation approach, the quality of evaluation, and clarity. Evaluation expectations are demanding in a specific way: because an adaptation mechanism can look impressive on a single favourable scenario, reviewers look for comparison against sensible baselines, discussion of overhead and failure cases, sensitivity to the assumptions built into a scenario, and, where possible, use of a recognised exemplar or a realistic application. A paper that proposes a loop, a language or a framework without evidence about how adaptation behaves when conditions deteriorate is unlikely to satisfy a SEAMS committee.

Page limits, categories, deadlines, formatting templates and any double-anonymous requirement are set per edition and are not publicly disclosed as permanent series-level rules; they change as the co-located conference's policies change. No standing acceptance rate for SEAMS could be verified from an official series-level source, and none is quoted here — acceptance rates quoted on third-party sites for individual years should be treated as unverified unless the edition's own chairs published them. Accepted papers are presented at the meeting and published in that edition's proceedings, with digital library records through ACM or IEEE depending on the edition's publishing arrangement. That variation explains a puzzle in the bibliographic record: a researcher may find one year's SEAMS papers in the ACM Digital Library and another year's in IEEE Xplore. Both can be correct. Always cite the specific edition you used, with its year, place and digital object identifier where one is given.

Indexing, metrics and citation practice

SEAMS proceedings are indexed as conference proceedings in the major computer science bibliographies, including DBLP, which maintains edition-by-edition records, and in Scopus, from which the JournalsWorld record's SJR value of 0.371 and H-index of 25 derive. Those stored figures describe the proceedings series as indexed at the snapshot used for the record; they are not a journal impact factor, and they should not be compared directly with journal quartiles. The record's quartile field is blank in its template, which is appropriate for a proceedings title of this kind. For an individual researcher, the more meaningful signals are the paper's own citations, whether later work reimplemented or compared against it, and whether it influenced a community exemplar, a survey or a Dagstuhl seminar roadmap. Surveys of software engineering for self-adaptive systems routinely use SEAMS, together with the related Dagstuhl seminars on engineering self-adaptive systems, as evidence of how the field matured from early autonomic computing ideas into a recognised software engineering discipline.

Citation hygiene matters because the title changed. A 2008 paper may be cited as appearing in the ICSE Workshop on Software Engineering for Adaptive and Self-Managing Systems, a 2013 paper as appearing in the International Symposium of the same name, and a 2020 paper under IEEE/ACM symposium wording. These are editions of the same SEAMS series, not rival conferences. Conversely, SEAMS should not be confused with general autonomic computing conferences, with ICSE itself, or with workshops on self-adaptive systems held inside other conferences in a single year. The tests are the SEAMS name, continuity of the community and programme leadership, and an edition number or year consistent with the sequence that began in Shanghai in 2006.

Who should attend and read SEAMS

The proceedings are essential reading for software engineering researchers working on self-adaptation, runtime models, adaptive architectures and assurance of evolving systems, and for researchers in cloud, services, robotics and cyber-physical systems who need adaptation mechanisms that can be engineered rather than improvised. Industry engineers concerned with automated operations, self-healing infrastructure and policy-driven management will find the applied papers and exemplars useful, although the venue's centre of gravity is research rather than product practice. New entrants to the field should begin with the survey and roadmap literature that SEAMS authors produced — notably the community synthesis on engineering self-adaptive systems through feedback loops — and then read forward through the proceedings editions that address their own domain.

Frequently asked questions

Is SEAMS a journal? No. It is a conference — first an ICSE workshop, now an international symposium — and this record describes its proceedings series.

What does SEAMS stand for? Software Engineering for Adaptive and Self-Managing Systems.

When and where did it begin? In 2006, as an ICSE workshop in Shanghai, consolidating earlier self-healing, autonomic and dependable-architecture workshops held at ICSE and FSE since 2002.

What are its ISSNs in this record? 2156-7891 and 2157-2305. Individual editions also carry ISBNs, and individual papers carry DOIs through the publishing digital library for that edition.

Who publishes the proceedings? The JournalsWorld record names IEEE Computer Society, United States. Historically, editions have appeared through IEEE and through ACM, so check the specific edition's record for its publisher and digital library.

What are its stored metrics? H-index 25 and SJR 0.371 in the JournalsWorld record. Those are proceedings-level, year-dependent figures, not a journal impact factor.

How do I submit? Through the call for papers of a specific future edition, by its deadline and in its stated categories. There is no rolling journal submission route and no article processing charge structure of the kind used by open access journals; conference registration and presentation requirements are set per edition.

Reading a SEAMS paper well

A SEAMS paper repays a particular kind of reading. Start with the adaptation problem rather than the technique: what changes in the environment or the system, how quickly must the system respond, and what goes wrong if it does not adapt at all? Then identify the knowledge the system keeps at run time — its models of itself, its goals and its surroundings — because the quality of adaptation can never exceed the quality of that knowledge. Next, examine the decision rule: is it a utility calculation, a set of policies, a planner, a learned model, or a combination of those, and what guarantees, if any, accompany it? Finally, read the evaluation against the claims, looking for overhead measurements, for scenarios in which adaptation was harmful or unnecessary, and for a discussion of threats to validity. Papers that survive that reading — clear problem, explicit knowledge, defensible decision rule, honest evaluation — are the papers the community tends to build on, cite in surveys, and reuse in exemplars. Papers that blur any of those four elements may still contain a useful prototype or dataset, and SEAMS artifact and short-paper formats give such work a legitimate home, provided the paper itself is candid about what has and has not been shown. Doctoral and early-career researchers in particular use SEAMS in this formative way, presenting a problem framing and an initial mechanism to a room of specialists who will test the framing harder than any journal reviewer, and returning in later years with the mature version. That developmental function, alongside its archival proceedings, is a large part of why a workshop founded in 2006 could grow into a symposium without losing its community.

Verification note

This profile is based on the JournalsWorld record for post 405888 (IEEE Computer Society, United States; ISSNs 2156-7891 and 2157-2305; stored H-index 25, SJR 0.371), on the SEAMS community and edition pages listing editions from 2006 in Shanghai onward, on the published tables of contents for the 2007, 2009 and 2019 editions, on DBLP-style proceedings records for 2013 and 2014, and on survey literature describing the consolidation of predecessor workshops in 2006. The structural correction to the original template is the same in kind as for the other proceedings in this batch: the template described a journal, whereas the verified entity is a conference proceedings series whose name later changed from workshop to symposium.