This is the Artifacts namespace. There are no child pages: what belongs here is the measurement-specific release problem, not a general lecture on open science. A PhD student who has just finished a crawl lands here from start and from Website selection, and needs to know what to put in the artefact, where to put it so a 2026 Available badge will accept it, and what to do when the data has people in it.

• Here: what a measurement artefact has to contain; which hosts the 2026 calls accept; the corpus rates; what to withhold and how to say so.
• Study preregistration: the thing a badge is not. 3,321 papers release a URL; 15 preregistered. They answer different questions.
• Ethics: IRB, residential proxies, crawler identification. This page only covers the release decision once the data exists.
• Website selection / Sampling / Longitudinal: pinning the list. Publishing the pinned list is the artefact step those pages point at.
• Crawler / OpenWPM: pinning the browser binary. Archive the pin with the artefact.

• Demir et al., TheWebConf 2022 [1Demir, Nurullah; Große-Kampmann, Matteo; Urban, Tobias; Wressnegger, Christian; Holz, Thorsten; Pohlmann, Norbert (2022): "Reproducibility and Replicability of Web Measurement Studies", in: Proceedings of the ACM Web Conference. (DOI)] — 117 web-measurement papers coded against 18 reproducibility criteria, plus a 4.5-million-page experiment that varies the setup on purpose. It is the checklist this page does not need to reinvent, and the evidence that the configuration you did not write down moves the number.
• Your venue's live artifact-evaluation page, not last year's. The 2026 calls disagree with each other (table below) and with 2024. The corpus cannot tell you which badge is current; the call can.
• The USENIX 2026 badges page if you are submitting there. Available is no longer optional, and GitHub is no longer enough.

A measurement artefact is not “the code”. The papers that cannot be re-run are missing the sample, the pins, and the per-site outcomes. Demir et al. [1Demir, Nurullah; Große-Kampmann, Matteo; Urban, Tobias; Wressnegger, Christian; Holz, Thorsten; Pohlmann, Norbert (2022): "Reproducibility and Replicability of Web Measurement Studies", in: Proceedings of the ACM Web Conference. (DOI)] coded 18 criteria; configuration details fail often (crawler adjustments, location, browser), and so does the repository: C9 “crawler publicly available” was omitted by 67% of their 117 papers, and only 24% made results openly available.

For a crawl, the minimum that lets someone else regenerate your headline number — or re-analyse it with a clustered standard error, see Hypothesis testing — is:

A system paper that evaluated a defence on a testbed has a different minimum (the implementation, the configs, the test vectors). This page is for the student who crawled.

Figures below are from scripts/report_artifacts.mjs over the 5,859-paper extraction (2010–2026, CCS, IMC, NDSS, PETS, USENIX Security, TheWebConf, IEEE S&P). “Own-link” means at least one artifacts.links[] entry with belongsToAuthors === true. That is the same definition as Corpus's “Releases an artifact link” column.

Dropping the 251 posters and ≤4-page records moves the all-papers rate 56.7% → 58.4% (3,277 / 5,608). The gap this page is about is not a poster artefact.

Availability is a different question from a URL. Among empirical papers, 2,890 of 5,118 (56.5%) state an availability other than none-mentioned / no record — OVERVIEW.md's artifacts.availability row. 2,845 papers in the whole corpus are labelled public; 54 of those have no authors'-own URL (they released “as part of scamper”, a WINE dataset id, a citation). 530 papers have an authors'-own URL and are not labelled public (promised, restricted, on-request, withheld, or the extractor stored the URL under none-mentioned). The headline uses the URL.

Paper-counted kinds, of the 3,321 with an authors'-own URL. A paper with source-code and a dataset is in both rows; the kinds sum to 4,199.

The 12 preregistration-kind links are the schema signal. Study preregistration's full-text sweep finds 15 papers that preregistered; do not publish 12 as the preregistration count. A venue badge is not a preregistration.

Hostnames folded into families (scripts/host_fold.mjs; residue printed on the provenance page). A paper with GitHub and Zenodo is in both rows. Shares are of 3,321.

Union of archival families: 622 / 3,321 = 18.7%. 59 papers (1.8%) give a shortener as the artifact URL (bit.ly, tinyurl, goo.gl). That URL is not an artifact; it is a redirect the owner can retarget. 667 distinct hosts did not match a family — almost all lab and university project pages. GitHub remains the default; the archival hosts are the minority, and they are new.

Own-link share of all papers that year, then GitHub / Zenodo share of the own-link papers. 2025–2026 starred: CCS/IMC 2026 unheld; IEEE S&P/WWW 2026 incompletely selected.

GitHub's share of own-link papers peaked in 2022 (69.2%) and dropped in 2025 (53.7%) as Zenodo rose. That is a venue-policy effect, not a GitHub decline in absolute terms: 2025 still has 318 GitHub papers, more than 2022's 234.

USENIX is the venue that currently requires an open-science appendix, and it is the venue with the highest own-link rate (69.5%) and more than half of the Zenodo mass (233 of 436). PETS, which has run optional artifact evaluation for years, has 5 Zenodo papers. Optional AE does not produce archival deposits; a mandatory Available badge does.

Checked 2026-08-27 against each call. Training data is stale here by construction. scripts/external_checks_artifacts.sh re-fetches these pages.

The schema's artifacts.badge field is non-null for 15 of 5,859 papers (0.3%). That is not a badge-award rate. Papers almost never write the badge name in a sentence the extractor can attach. A full-text sweep for “artifact evaluation” hits 59 papers — still an upper bound, still not an award rate. Do not use this corpus to say how often a venue awards badges. Use the venue's own AE results page, as NDSS 2026 publishes.

GitHub is the right place to develop. It is the wrong place to cite. The default branch moves; a force-push rewrites history; a renamed account 404s the camera-ready URL. The 2026 Available calls noticed.

The pattern that satisfies every row of the table above:

1. Develop on GitHub.
2. At camera-ready, make a release (a tag, not “latest”).
3. Push that release to Zenodo (the GitHub–Zenodo integration mints a DOI per release). You get a version DOI (this snapshot) and a concept DOI (the record that will always resolve to the newest version). Cite the version DOI in the paper.
4. Leave the GitHub URL in the README so people can file issues. It is a pointer, not the archive.

OSF, Figshare, Dryad, institutional Dataverse, 4TU.ResearchData and Software Heritage also issue DOIs. Zenodo is the one the security AE pages name first, and it is the one 436 papers in this corpus already use.

anonymous.4open.science (36 papers) is a double-blind stand-in for GitHub during review. It is not the camera-ready archive. Swap it for a DOI before the Available check.

USENIX 2026's open-science appendix requires you to explain a missing artifact. “We do not release the data” without a reason is no longer a complete sentence at that venue. The reasons in this corpus are real, and they cluster.

People in the data. The reasons cluster. These are the sentences, not a typology we imposed.

Google's Right to be Forgotten paper cannot publish the URL list it classified, because the list is the sensitive object [4Bertram, Theo; Bursztein, Elie; Caro, Stephanie; Chao, Hubert; Feman, Rutledge Chin; Fleischer, Peter; Gustafsson, Albin; Hemerly, Jess; Hibbert, Chris; Invernizzi, Luca; Donnelly, Lanah Kammourieh; Ketover, Jason; Laefer, Jay; Nicholas, Paul; Niu, Yuan; Obhi, Harjinder; Price, David; Strait, Andrew; Thomas, Kurt; Verney, Al (2019): "Five Years of the Right to be Forgotten", in: Proceedings of the ACM SIGSAC Conference on Computer and Communications Security. (DOI)]. Lyons et al. logged identifiers off Android devices [3Lyons, Allan; Gamba, Julien; Shawaga, Austin; Reardon, Joel; Tapiador, Juan; Egelman, Serge; Vallina-Rodriguez, Narseo (2023): "Log: It's Big, It's Heavy, It's Filled with Personal Data! Measuring the Logging of Sensitive Information in the Android Ecosystem", in: Proceedings of the USENIX Security Symposium. (Link)]. Dey, Ding and Ross crawled Facebook profiles of high-school students [2Dey, Ratan; Ding, Yuan; Ross, Keith W. (2013): "Profiling high-school students with facebook: how online privacy laws can actually increase minors' risk", in: Proceedings of the ACM Internet Measurement Conference. (DOI)].

Operator NDAs and campus traffic. IMC is full of this: COVID-era mobile-operator traces, campus NetFlow, MNO handover logs, all under agreements that forbid re-sharing even for research. The paper can still release the analysis code, the schema, and synthetic or aggregated tables.

Dual-use. Kim et al. built LTEFuzz and wrote “A public release is not planned as LTEFuzz can be used for malicious purposes”; they planned a private release to carriers [5Kim, Hongil; Lee, Jiho; Lee, Eunkyu; Kim, Yongdae (2019): "Touching the Untouchables: Dynamic Security Analysis of the LTE Control Plane", in: Proceedings of the IEEE Symposium on Security and Privacy. (DOI)]. That is a different withholding from PII. The harm is the capability, not the dataset. See Ethics on dual-use.

Restricted access is a release. Pastrana et al.'s CrimeBB is available to academic researchers from the Cambridge Cybercrime Centre, not on a public URL [6Pastrana, Sergio; Thomas, Daniel R.; Hutchings, Alice; Clayton, Richard (2018): "CrimeBB: Enabling Cybercrime Research on Underground Forums at Scale", in: Proceedings of the ACM Web Conference. (DOI)]. Beverly et al. published seeds and targets and held the complete IPv6 traces under restricted distribution [7Beverly, Robert; Durairajan, Ramakrishnan; Plonka, David; Rohrer, Justin P. (2018): "In the IP of the Beholder: Strategies for Active IPv6 Topology Discovery", in: Proceedings of the ACM Internet Measurement Conference. (DOI)]. Both count as an artifact in the schema (restricted / a dataset URL). A reviewer can still get the data. A bit.ly link to a Google Drive folder that you might delete is worse than a named restricted archive.

What you can always release, even when the raw crawl cannot go out:

• The URL list (or the Tranco id plus the filter), unless the URLs themselves are the sensitive object (delisting requests, victim sites, the high-school Facebook study).
• The crawler, the pins, the classifier, the analysis scripts.
• Aggregates and per-bin counts that do not identify a site or a person.
• A data dictionary of what you deleted.

Thomas et al. crawled ad injection from inside Google and wrote “never sharing raw data outside of Google” [8Thomas, Kurt; Bursztein, Elie; Grier, Chris; Ho, Grant; Jagpal, Nav; Kapravelos, Alexandros; McCoy, Damon; Nappa, Antonio; Paxson, Vern; Pearce, Paul; Provos, Niels; Rajab, Moheeb Abu (2015): "Ad Injection at Scale: Assessing Deceptive Advertisement Modifications", in: Proceedings of the IEEE Symposium on Security and Privacy. (DOI)]. Only aggregates could be retained. That is a complete answer; “the data is sensitive” is not.

pin_artifact.py classifies a URL the way the 2026 Available calls do, and for a Zenodo record prints the version DOI, the concept DOI, the files and the licence. --require-doi exits 1 unless every URL is on a named archival host (Zenodo, OSF, Figshare, Dryad, Dataverse, 4TU, Software Heritage). A publisher DOI is not enough; a GitHub path that happens to contain 10.5281 is not Zenodo.

pin_artifact.py

#!/usr/bin/env python3
"""Is this artifact URL archival, or just a pointer at a mutable repo?
 
USENIX Security 2026's Artifacts Available badge refuses GitHub, GitLab and
personal pages: the archived copy has to be reachable via a long-term stable
reference or DOI. IEEE S&P 2026 and TheWebConf 2026 say the same for Available.
This script classifies a URL the way those calls do, and for a Zenodo record
prints the version DOI, the concept DOI, the files and the licence — so a
camera-ready statement can cite the record that was actually deposited, not
the living GitHub default branch.
 
A publisher DOI (ACM, IEEE, …) is not an archival artifact. --require-doi
accepts only named archives (Zenodo, OSF, Figshare, Dryad, Dataverse, 4TU,
Software Heritage) and their known DOI prefixes, not doi.org in general.
 
    uv run python pin_artifact.py https://doi.org/10.5281/zenodo.1421702
    uv run python pin_artifact.py --require-doi https://github.com/mozilla/OpenWPM
    uv run python pin_artifact.py --file urls.txt
 
Exit status is 1 if any URL is a shortener, is unparseable, is a Zenodo URL
without a record id, or (with --require-doi) is not on a named archival host.
No API key needed for Zenodo.
"""
from __future__ import annotations
 
import argparse
import json
import re
import sys
import urllib.error
import urllib.parse
import urllib.request
 
UA = "measuretheweb-pin-artifact/1.0 (https://measuretheweb.org/artifacts)"
 
# Ordered: first match wins. Keep in sync with scripts/host_fold.mjs.
# (family, permanence, host-or-path test)
DOI_ARCHIVAL = {
    "10.5281": "Zenodo",
    "10.17605": "OSF",
    "10.6084": "Figshare",
    "10.5061": "Dryad",
    "10.7910": "Harvard Dataverse",
    "10.4121": "4TU.ResearchData",
}
 
ARCHIVAL_HOSTS = {
    "zenodo.org": "Zenodo",
    "osf.io": "OSF",
    "figshare.com": "Figshare",
    "datadryad.org": "Dryad",
    "dryad.org": "Dryad",
    "dataverse.harvard.edu": "Harvard Dataverse",
    "data.4tu.nl": "4TU.ResearchData",
    "archive.softwareheritage.org": "Software Heritage",
}
 
MUTABLE_HOSTS = {
    "github.com": "GitHub",
    "gist.github.com": "GitHub",
    "githubusercontent.com": "GitHub",
    "gitlab.com": "GitLab",
    "bitbucket.org": "Bitbucket",
    "huggingface.co": "Hugging Face",
    "drive.google.com": "Google Drive / Docs",
    "docs.google.com": "Google Drive / Docs",
    "dropbox.com": "Dropbox",
}
 
SHORTENERS = {
    "bit.ly",
    "tinyurl.com",
    "goo.gl",
    "t.ly",
    "ow.ly",
    "is.gd",
    "cutt.ly",
    "rebrand.ly",
    "shorturl.at",
}
DOI_HOSTS = {"doi.org", "dx.doi.org"}
 
ZENODO_ID_RE = re.compile(r"(?:zenodo\.org/(?:records|record)/|10\.5281/zenodo\.)(\d+)", re.I)
 
 
def parse(url: str) -> tuple[str, str, str]:
    raw = url.strip()
    if raw == "":
        raise ValueError("empty URL")
    m = re.match(r"^doi:\s*(10\.\S+)", raw, re.I)
    if m:
        return "doi.org", "/" + m.group(1), raw
    with_scheme = raw if re.match(r"^[a-z][a-z0-9+.-]*://", raw, re.I) else "https://" + raw
    parts = urllib.parse.urlparse(with_scheme.replace(" ", ""))
    host = parts.hostname
    if host is None or host == "":
        raise ValueError(f"no host in {raw!r}")
    if host.startswith("www."):
        host = host[4:]
    return host.lower(), parts.path, raw
 
 
def doi_prefix(path: str) -> str | None:
    m = re.search(r"(10\.\d+)", path)
    if m is None:
        return None
    return m.group(1)
 
 
def classify(url: str) -> tuple[str, str]:
    """Return (family, permanence).
 
    DOI-prefix matching applies only to doi.org / dx.doi.org / doi: URLs.
    A GitHub path that happens to contain 10.5281 is not Zenodo.
    A generic publisher DOI is not archival.
    """
    host, path, raw = parse(url)
    if host in SHORTENERS:
        return "URL shortener", "shortener"
    is_doi = host in DOI_HOSTS
    if is_doi:
        pref = doi_prefix(path) or doi_prefix(raw)
        if pref is not None and pref in DOI_ARCHIVAL:
            return DOI_ARCHIVAL[pref], "archival"
        if pref == "10.48550":
            return "arXiv", "publisher-page"
        return "DOI resolver (unknown archive)", "unknown"
    if host in ARCHIVAL_HOSTS:
        family = ARCHIVAL_HOSTS[host]
        if family == "Zenodo" and zenodo_id(raw) is None:
            return "Zenodo (no record id)", "unknown"
        return family, "archival"
    if host.endswith(".osf.io"):
        return "OSF", "archival"
    if host.endswith(".figshare.com"):
        return "Figshare", "archival"
    if host.endswith(".github.io") or host.endswith(".githubusercontent.com"):
        return "GitHub", "mutable-repo"
    if host in MUTABLE_HOSTS:
        return MUTABLE_HOSTS[host], "mutable-repo"
    if host.endswith(".dropbox.com"):
        return "Dropbox", "mutable-repo"
    if host == "sites.google.com":
        return "Google Sites (project page)", "project-page"
    if host == "anonymous.4open.science":
        return "anonymous.4open.science (double-blind GitHub)", "anonymized"
    if host == "arxiv.org":
        return "arXiv", "publisher-page"
    if host == "usenix.org" or host.endswith(".usenix.org"):
        return "USENIX paper page", "publisher-page"
    return f"unmapped host {host}", "unknown"
 
 
def get_json(url: str) -> dict:
    req = urllib.request.Request(url, headers={"Accept": "application/json", "User-Agent": UA})
    with urllib.request.urlopen(req, timeout=30) as r:
        return json.load(r)
 
 
def zenodo_id(url: str) -> str | None:
    m = ZENODO_ID_RE.search(url)
    if m is None:
        return None
    return m.group(1)
 
 
def describe_zenodo(url: str) -> str:
    rec_id = zenodo_id(url)
    if rec_id is None:
        return "could not parse a Zenodo record id"
    try:
        rec = get_json(f"https://zenodo.org/api/records/{rec_id}")
    except urllib.error.HTTPError as e:
        return f"Zenodo API HTTP {e.code} for record {rec_id}"
    md = rec["metadata"]
    files = rec["files"]
    version_doi = md["doi"]
    concept = md["relations"]["version"][0]["parent"]["pid_value"]
    created = rec["created"][:10]
    license_id = md["license"]["id"]
    titles = ", ".join(f["key"] + f" ({f['size']} bytes)" for f in files[:8])
    more = "" if len(files) <= 8 else f" … +{len(files) - 8} files"
    return (
        f"record {rec_id}; version DOI 10.5281/zenodo.{rec_id} ({version_doi}); "
        f"concept DOI 10.5281/zenodo.{concept}; created {created}; "
        f"licence {license_id}; {len(files)} files: {titles}{more}"
    )
 
 
def main() -> int:
    ap = argparse.ArgumentParser()
    ap.add_argument("urls", nargs="*")
    ap.add_argument("--file", help="one URL per line")
    ap.add_argument(
        "--require-doi",
        action="store_true",
        help="exit 1 unless every URL is on a named archival host (Zenodo/OSF/Figshare/Dryad/Dataverse/4TU/Software Heritage). A publisher DOI is not enough.",
    )
    args = ap.parse_args()
    urls = list(args.urls)
    if args.file:
        urls += [line.strip() for line in open(args.file) if line.strip() != ""]
    if len(urls) == 0:
        ap.error("give at least one URL")
 
    bad = 0
    for url in urls:
        try:
            family, permanence = classify(url)
        except ValueError as e:
            print(f"{'UNPARSEABLE':<12} {url}  ({e})")
            bad += 1
            continue
        extra = ""
        if family == "Zenodo":
            desc = describe_zenodo(url)
            extra = "  |  " + desc
            if desc.startswith("could not parse") or desc.startswith("Zenodo API HTTP"):
                bad += 1
        print(f"{permanence:<12} {family:<44} {url}{extra}")
        if permanence == "shortener":
            bad += 1
        if permanence == "unknown" and family.startswith("Zenodo"):
            bad += 1
        if args.require_doi and permanence != "archival":
            bad += 1
    return 1 if bad else 0
 
 
if __name__ == "__main__":
    raise SystemExit(main())

Live run, 2026-08-27. The Zenodo record is an arbitrary deposited snapshot, used here to show the API fields, not as a measurement artefact to imitate. OpenWPM's GitHub URL is the counterexample: it is the right development URL and it fails an Available check.

uv run python pin_artifact.py https://github.com/mozilla/OpenWPM https://doi.org/10.5281/zenodo.1421702

mutable-repo GitHub                                       https://github.com/mozilla/OpenWPM
archival     Zenodo                                       https://doi.org/10.5281/zenodo.1421702  |  record 1421702; version DOI 10.5281/zenodo.1421702 (10.5281/zenodo.1421702); concept DOI 10.5281/zenodo.1421701; created 2018-09-19; licence other-open; 1 files: Eichhoernchen/monero-digging-paper-v1.zip (20788 bytes)

uv run python pin_artifact.py –require-doi https://github.com/mozilla/OpenWPM (exits 1)

mutable-repo GitHub                                       https://github.com/mozilla/OpenWPM

Cite OpenWPM itself as [9Englehardt, Steven; Narayanan, Arvind (2016): "Online Tracking: A 1-million-site Measurement and Analysis", in: Proceedings of the 2016 ACM SIGSAC Conference on Computer and Communications Security, pp. 1388–1401. Association for Computing Machinery, New York, NY, USA. (DOI) (Link)]; the GitHub URL above is the project, not a 2016 snapshot.

1. A DOI, in the paper, next to the GitHub URL. The GitHub URL is for issues. The DOI is the citation.
2. The version DOI, not the concept DOI, if you want the reader to get this snapshot.
3. The list, the pins, the per-site table, not only the crawler.
4. The licence on the deposit. Zenodo will not let you mint a DOI without one; GitHub will.
5. If you withheld: the reason, and what you released instead.
6. No shortener. 59 papers in this corpus did that.
7. No Google Drive / Dropbox as the archival copy. 41 + 13 papers. Fine as a working folder; not a camera-ready artifact.
8. anonymous.4open.science is for review. Replace it.

• Study preregistration — a badge does not pin your analysis plan.
• Ethics — the review and the harm analysis that decide whether you can release.
• Website selection — publish the list you actually fetched.
• Sampling / Longitudinal — the pins a later wave needs.
• Crawler — the browser binary belongs in the artefact.
• Hypothesis testing / Pvalue corrections — per-site outcomes are what make a re-analysis possible.
• Corpus — seven venues, 2010–2026, 5,859 papers.

Every corpus figure above comes from scripts/report_artifacts.mjs over data/extract/run1 (5,859 papers). Limits that carry:

• Seven venues only. CHI, SOUPS, ACSAC, EuroS&P, RAID, AsiaCCS are absent. Artifact-evaluation culture at those venues is not in these rates.
• “Own-link” is a URL the extractor stored as the authors'. A paper that says “code available on request” with no URL is not in the 3,321. A paper that prints someone else's GitHub is not either (belongsToAuthors === false).
• Hosts are folded. GitHub includes gist, raw.githubusercontent.com and *.github.io. Zenodo includes doi.org/10.5281 and the unschemed doi:10.5281/zenodo.N. The residue (667 hosts) is on the provenance page. Exact-string “github.com” undercounts.
• artifacts.badge is not a badge-award rate. 15 papers. Use the venue AE results.
• 2025–2026 are provisional. Own-link 76.9% / 75.9% in those years will move when the missing venue-years land.
• data/extract/artifact_links.jsonl is the old 4,322-paper corpus. This page does not use it. OVERVIEW.md still points at it; it is stale.

The complete query log, the report with its unedited output, the host fold and its residue, every quote with its verification verdict, and every external source with how it was verified are on artifacts. Corpus-wide caveats are on Corpus.

[1]

Demir, Nurullah; Große-Kampmann, Matteo; Urban, Tobias; Wressnegger, Christian; Holz, Thorsten; Pohlmann, Norbert (2022): "Reproducibility and Replicability of Web Measurement Studies", in: Proceedings of the ACM Web Conference. (DOI)

[2]

Dey, Ratan; Ding, Yuan; Ross, Keith W. (2013): "Profiling high-school students with facebook: how online privacy laws can actually increase minors' risk", in: Proceedings of the ACM Internet Measurement Conference. (DOI)

[3]

Lyons, Allan; Gamba, Julien; Shawaga, Austin; Reardon, Joel; Tapiador, Juan; Egelman, Serge; Vallina-Rodriguez, Narseo (2023): "Log: It's Big, It's Heavy, It's Filled with Personal Data! Measuring the Logging of Sensitive Information in the Android Ecosystem", in: Proceedings of the USENIX Security Symposium. (Link)

[4]

Bertram, Theo; Bursztein, Elie; Caro, Stephanie; Chao, Hubert; Feman, Rutledge Chin; Fleischer, Peter; Gustafsson, Albin; Hemerly, Jess; Hibbert, Chris; Invernizzi, Luca; Donnelly, Lanah Kammourieh; Ketover, Jason; Laefer, Jay; Nicholas, Paul; Niu, Yuan; Obhi, Harjinder; Price, David; Strait, Andrew; Thomas, Kurt; Verney, Al (2019): "Five Years of the Right to be Forgotten", in: Proceedings of the ACM SIGSAC Conference on Computer and Communications Security. (DOI)

[5]

Kim, Hongil; Lee, Jiho; Lee, Eunkyu; Kim, Yongdae (2019): "Touching the Untouchables: Dynamic Security Analysis of the LTE Control Plane", in: Proceedings of the IEEE Symposium on Security and Privacy. (DOI)

[6]

Pastrana, Sergio; Thomas, Daniel R.; Hutchings, Alice; Clayton, Richard (2018): "CrimeBB: Enabling Cybercrime Research on Underground Forums at Scale", in: Proceedings of the ACM Web Conference. (DOI)

[7]

Beverly, Robert; Durairajan, Ramakrishnan; Plonka, David; Rohrer, Justin P. (2018): "In the IP of the Beholder: Strategies for Active IPv6 Topology Discovery", in: Proceedings of the ACM Internet Measurement Conference. (DOI)

[8]

Thomas, Kurt; Bursztein, Elie; Grier, Chris; Ho, Grant; Jagpal, Nav; Kapravelos, Alexandros; McCoy, Damon; Nappa, Antonio; Paxson, Vern; Pearce, Paul; Provos, Niels; Rajab, Moheeb Abu (2015): "Ad Injection at Scale: Assessing Deceptive Advertisement Modifications", in: Proceedings of the IEEE Symposium on Security and Privacy. (DOI)

[9]

Englehardt, Steven; Narayanan, Arvind (2016): "Online Tracking: A 1-million-site Measurement and Analysis", in: Proceedings of the 2016 ACM SIGSAC Conference on Computer and Communications Security, pp. 1388–1401. Association for Computing Machinery, New York, NY, USA. (DOI) (Link)