<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Keerthana Purushotham on Medium]]></title>
        <description><![CDATA[Stories by Keerthana Purushotham on Medium]]></description>
        <link>https://medium.com/@keerthanapurushotham?source=rss-ab7ae3394a5c------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*HtRJCTzEVphQt0wLCOI8LQ.jpeg</url>
            <title>Stories by Keerthana Purushotham on Medium</title>
            <link>https://medium.com/@keerthanapurushotham?source=rss-ab7ae3394a5c------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 14:07:57 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@keerthanapurushotham/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Accuracy Is Not Enough: Confusion Matrix Metrics That Actually Work in CVE Impact Prediction]]></title>
            <link>https://medium.com/@keerthanapurushotham/accuracy-is-not-enough-confusion-matrix-metrics-that-actually-work-in-cve-impact-prediction-d4bafd9cec1b?source=rss-ab7ae3394a5c------2</link>
            <guid isPermaLink="false">https://medium.com/p/d4bafd9cec1b</guid>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[vulnerability-management]]></category>
            <category><![CDATA[statistics]]></category>
            <category><![CDATA[nlp]]></category>
            <category><![CDATA[linux]]></category>
            <dc:creator><![CDATA[Keerthana Purushotham]]></dc:creator>
            <pubDate>Fri, 19 Sep 2025 21:43:25 GMT</pubDate>
            <atom:updated>2025-12-03T20:27:32.172Z</atom:updated>
            <content:encoded><![CDATA[<h3>Accuracy Is Not Enough</h3><h4>Confusion Matrix Metrics That Actually Work in CVE Impact Prediction</h4><blockquote>Copyright © 2025 Keerthana Purushotham &lt;keep.consult@proton.me&gt;.<br>Licensed under the GNU AGPL v3. See <a href="https://github.com/keerthanap8898/Accuracy-is-Not-Enough-in-Cybersecurity/blob/main/LICENSE"><strong>LICENSE</strong></a> for details.<br>To cite, scroll to the bottom, go to <a href="https://doi.org/10.5281/zenodo.17438182"><strong>Zenodo</strong></a> ( <strong>doi.org/10.5281/zenodo.17438182</strong> ) or <a href="https://github.com/keerthanap8898/Accuracy-is-Not-Enough-in-Cybersecurity/blob/main/README.md#githubcomkeerthanap8898accuracy-is-not-enough-in-cybersecuritytreemaincite_this_article"><strong>Github</strong></a>.</blockquote><p>In cybersecurity, mapping vulnerabilities (CVEs) across Linux distributions is not just classification — it’s risk control. Whether python-requests in Debian matches python3-requests in Red Hat can mean the difference between a patched system and an exploitable one</p><h4><strong>Why does this matter?</strong></h4><p>False negatives (missed vulnerabilities) leave blind spots. False positives (over-flagged safe packages) burn developer time and erode trust. Accuracy hides both problems.</p><p><strong>The solution: </strong>treat the <strong>confusion matrix as a toolbox</strong>, and climb through its metrics step by step.</p><figure><img alt="Leveled Confusion Matrix Metrics Framework (L0–L6)" src="https://cdn-images-1.medium.com/max/1024/1*hgEVXUaiPipLnockutDMjQ.png" /><figcaption>Leveled Confusion Matrix Metrics Framework (L0–L6)</figcaption></figure><h4>Trade-Offs:</h4><p><strong>👉🏽 False Negatives</strong></p><p>The cost of not only missing the goal but also causes loss that can end in catastrophe. (e.g. market anomaly/ 0-day CVE miss, sepsis undetected until too late, or a missed satellite collision that wipes out a B$ asset. That’s why systems go heavy on redundancy, richer signals &amp; constant error analysis to catch what matters most.)</p><p><strong>👉🏽 False Positives</strong></p><p>Less risk, but they kill trust/efficiency. Too many false alarms=alert fatigue, labor etc(eg:trade-entries falsely flagged as fraud, too many SRE TTs,AV braking on empty roads, false debris alarms[i.e fuel waste]- in space. Even justice systems suffer from facial recognition, to wrong risk scores disrupting real lives).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/986/1*ij253ryy2xQVrosGUgUk_A.png" /><figcaption>Domain specific examples</figcaption></figure><figure><img alt="FN vs FP Tradeoffs in Risk, visualized across various industries" src="https://cdn-images-1.medium.com/max/1024/1*YuKoNsdPwZnUgfLdTb3DRA.png" /><figcaption>FN vs FP Tradeoffs in Risk, visualized across various industries</figcaption></figure><h4>Level 0: Core Counts (Primitives)</h4><p>The foundation of all metrics:</p><ul><li><strong><em>TP (True Positive)</em>:</strong> Vulnerable package correctly flagged</li><li><strong><em>TN (True Negative)</em>:</strong> Safe package correctly ignored</li><li><strong><em>FP (False Positive)</em>:</strong> Safe package wrongly flagged</li><li><strong><em>FN (False Negative)</em>:</strong> Vulnerable package missed</li></ul><h4>Level 1: Class Rates (per positive/negative)</h4><p>⭆ <strong><em>Recall</em>: </strong>Prioritizes catching vulnerabilities. In security, recall is mission-critical.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/224/1*3tpeXN0SMJYeMzXrcsa8YA.png" /></figure><p>⭆ <strong><em>Specificity (TNR)</em>: </strong>Ensures safe packages aren’t over-flagged.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/235/1*urndGaU5ffxzsYa-2B8E9w.png" /></figure><p>Complements:</p><p>⇒ <strong><em>FNR</em> = 1 — Recall</strong></p><p>⇒ <strong><em>FPR</em> = 1 — Specificity</strong></p><p><em>Security takeaway: FNR is the costliest error. FPR is the loudest error.</em></p><h4>Level 2: Predictive Values (per outcome)</h4><p>⭆ <strong><em>Precision (PPV)</em>: </strong>How often a flagged package is truly vulnerable.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/214/1*RlxMYAQeWlMQFrdVeCEWfQ.png" /></figure><p>⭆ <strong><em>Negative Predictive Value (NPV)</em>: </strong>Measures how often a safe-labeled package is truly safe.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/181/1*7TxKOnXZ8Nlp6QWVrkLWgA.png" /></figure><p>Complements:</p><p>⇒ <strong><em>FDR</em> = 1 — Precision</strong></p><p>⇒ <strong><em>FOR</em> = 1 — NPV</strong></p><p><em>Security takeaway: FOR is underused but vital — it tells you the risk of trusting a “safe” prediction.</em></p><h4>Level 3: Likelihood Ratios (decision-level odds)</h4><p>⭆ <strong><em>Positive Likelihood Ratio (PLR)</em>: </strong>How much a positive result raises odds of vulnerability.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/147/1*pO2xo6c6R3rtIImsdxL3aQ.png" /></figure><p>⭆ <strong><em>Negative Likelihood Ratio (NLR)</em>: </strong>How much a negative result lowers odds of vulnerability.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/193/1*hZbOtLAlitRVRc6vVc1C7g.png" /></figure><p><em>&gt; Security takeaway: PLR = alert trustworthiness. NLR = safe decision trustworthiness.</em></p><h4>Level 4: Balanced Metrics (threshold trade-offs)</h4><p>⭆ <strong><em>F1-Positive</em>: </strong>Balances catching vulnerabilities with avoiding wasted alerts.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/274/1*EPW7Npplp5m3TRcJ-fIRCQ.png" /></figure><p>⭆ <strong><em>F1-Negative</em>: </strong>Balances suppressing false alarms with maintaining trust.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/278/1*SUqsjZcDVtUvE2V7mFWjjg.png" /></figure><p>⭆ <strong><em>Balanced Accuracy (BA)</em>: </strong>Reduces bias in imbalanced datasets.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/265/1*DrcXE9NUIg0qXtDzbC97QA.png" /></figure><p>⭆ <strong><em>Youden’s J</em>: </strong>Quick threshold selector maximizing separation from randomness.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/258/1*zL19_vcoAKdfjnKKfBYLSA.png" /></figure><h4>Level 5: Agreement and Global Correlation</h4><p>⭆ <strong><em>Cohen’s Kappa (κ)</em>: </strong>Useful for proving reliability across noisy datasets.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/133/1*fcoofsVoLOvd2c95l4WR7w.png" /></figure><p>where:</p><p>⇒ <strong>pₒ = </strong>observed agreement</p><p>⇒ <strong>pₑ = </strong>expected agreement by chance</p><p>⭆ <strong><em>Matthews Correlation Coefficient (MCC)</em>: </strong>Gold-standard robustness metric. Stable under imbalance and interpretable across security, bioinformatics, and finance.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/478/1*rPfTgNI9Q6-CIHQ6CH3LIQ.png" /></figure><h4>Level 6: Baseline Snapshot</h4><p>⭆ <strong><em>Accuracy</em>: </strong>A quick overall number — but misleading when vulnerabilities are rare.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/304/1*skR0CN_mM2_jG_TsU-kPUQ.png" /></figure><blockquote><em>Security takeaway: Accuracy is for dashboards, not risk decisions.</em></blockquote><h3>Practical Recommendations</h3><ul><li>Lead with <strong>Recall and FNR</strong> — <em>because misses are catastrophic</em></li><li>Track <strong>FOR and NLR</strong> — <em>because “safe” predictions must be safe</em></li><li>Use <strong>Balanced Accuracy and Youden’s J</strong> — <em>to set thresholds</em></li><li>Report <strong>MCC and F1</strong> — <em>for global robustness</em></li><li>Treat <strong>Accuracy</strong> — <em>as a top-line sanity check only</em></li></ul><p>Both matter: FN kill you outright, FP bleed you slowly. Domain awareness is 🔑.</p><h4>Glossary of Acronyms</h4><blockquote><strong>1. CVE [Common Vulnerabilities and Exposures]: </strong>standardized list of publicly disclosed cybersecurity vulnerabilities. <br><strong>2. TP [True Positive]:</strong> vulnerable package correctly identified as vulnerable. <br><strong>3. TN [True Negative]: </strong>safe package correctly identified as safe. <br><strong>4. FP [False Positive]:</strong> safe package incorrectly flagged as vulnerable. <br><strong>5. FN [False Negative]:</strong> vulnerable package incorrectly classified as safe (the most costly mistake in security). <br><strong>6. TPR [True Positive Rate / Recall / Sensitivity]:</strong> proportion of vulnerabilities correctly identified. <br><strong>7. TNR [True Negative Rate / Specificity]:</strong> proportion of safe packages correctly ignored. <br><strong>8. FNR [False Negative Rate]: </strong>proportion of vulnerabilities missed; complement of Recall. <br><strong>9. FPR [False Positive Rate]: </strong>proportion of safe packages incorrectly flagged; complement of Specificity. <br><strong>10. PPV [Positive Predictive Value / Precision]:</strong> proportion of flagged items that are truly vulnerable. <br><strong>11. NPV [Negative Predictive Value]:</strong> proportion of safe-labeled items that are truly safe. <br><strong>12. FDR [False Discovery Rate]: </strong>proportion of flagged positives that are actually false; complement of Precision. <br><strong>13. FOR [False Omission Rate]:</strong> proportion of negative predictions that are actually false; complement of NPV. <br><strong>14. PLR [Positive Likelihood Ratio]: </strong>how much more likely a positive prediction indicates vulnerability. <br><strong>15. NLR [Negative Likelihood Ratio]: </strong>how much less likely a negative prediction indicates vulnerability. <br><strong>16. BA [Balanced Accuracy]: </strong>average of Recall and Specificity, mitigating bias in imbalanced data. <br><strong>17. MCC [Matthews Correlation Coefficient]:</strong> correlation-style metric incorporating all four confusion matrix elements (TP, TN, FP, FN). <br><strong>18. κ [Cohen’s Kappa]: </strong>chance-corrected measure of agreement between predictions and ground truth.</blockquote><figure><img alt="(updated) Metrics table/cheat-sheet image" src="https://cdn-images-1.medium.com/max/1024/1*RglepN1K27XzXADc1yTrjQ.png" /><figcaption>(updated) Metrics table/cheat-sheet image</figcaption></figure><p>They’re all metrics that are easy to map with ingestion tools too — niche statistics you’d commonly see around ML models but mathematically speaking, they apply to any statistically, predictive black box.</p><h4>For example,</h4><p><strong>CVE-2023–48022</strong> is an epitome of how a disputed CVE turned into a false-negative, hiding an actively exploited bug downstream. Tracking False-Negative-Rate (FNR) or Negative-Likelihood-Ratio (NLR) would’ve flagged very early on that “safe/unaffected” wasn’t safe. Even MCC would’ve exposed the imbalance of missing rare but critical vulnerabilities.</p><p>The <a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/2023/48xxx/CVE-2023-48022.json">JSON for OSV template</a> has attributes that can be automated,<br>- CNA [disputed] → implying no urgent fix needed.<br>- CISA [affected] → treat as dangerous.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/537/1*LqjVKmff4QyZTI2gA78WMg.png" /><figcaption><strong>CVE-2023–48022 : </strong><a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/2023/48xxx/CVE-2023-48022.json">github.com/CVEProject/cvelistV5/blob/main/cves/2023/48xxx/CVE-2023-48022.json</a></figcaption></figure><p>That’s the FN suppressed incorrectly when it was being actively exploited.<br>- FNR (↑) spikes to 1 if an exploited vulnerability is missed.<br>- NLR (↑) the “safe/unaffected” label was untrustworthy.<br>- MCC (↓) correlation drops since a rare but critical positive was missed.</p><p><strong>But all this obviously depends on the timing of &amp; if exploitations/bugs in your tool are identifiable.</strong></p><p>To help visualize the dashboard, I plotted a sample and it would look something like this — [image attached below].</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*l9DrXh-K5vMHPiJ5q2EmIw.png" /><figcaption>The blue arrows along the MCC graph indicates decrease in overall prediction-mapping trust.</figcaption></figure><h4>Closing</h4><blockquote>Security &amp; threat impact prediction is not about being “mostly right”. It’s about being “right where it counts”.</blockquote><p>This leveled framework turns the confusion matrix into a decision-making map — one that reduces blind spots, manages operational noise, and builds confidence that predictions are aligned with real-world risk.</p><h3>Cite this article -</h3><p>( <a href="https://github.com/keerthanap8898/Accuracy-is-Not-Enough-in-Cybersecurity/blob/main/README.md#cite-this-article--"><em>github.com/keerthanap8898/Accuracy-is-Not-Enough-in-Cybersecurity</em> </a>)</p><h4><strong>Plain Text:</strong></h4><pre>```text/md<br>    Purushotham, K. (2025, September 19th). *Accuracy is not enough: Confusion matrix metrics that actually work in CVE impact prediction.* Zenodo - [DOI 10.5281/zenodo.17438182](https://doi.org/10.5281/zenodo.17438183) .<br>    <br>    Also available at:<br>    - Substack: https://keerthanapurushotham.substack.com/p/accuracy-is-not-enough-confusion<br>    - Medium: https://medium.com/@keerthanapurushotham/accuracy-is-not-enough-confusion-matrix-metrics-that-actually-work-in-cve-impact-prediction-d4bafd9cec1b  <br>    - GitHub: https://github.com/keerthanap8898/Accuracy-is-Not-Enough-in-Cybersecurity/tree/main<br>```</pre><h4>APA (7th Edition):</h4><pre>```text/md<br>    Purushotham, K. (2025, [September 19th]). Accuracy is not enough: Confusion matrix metrics that actually work in CVE impact prediction. DOI 10.5281/zenodo.17438182 - ( https://doi.org/10.5281/zenodo.17438183 ) .<br>    Also available at:<br>    - [Substack](https://keerthanapurushotham.substack.com/p/accuracy-is-not-enough-confusion)<br>    - [Medium](https://medium.com/@keerthanapurushotham/accuracy-is-not-enough-confusion-matrix-metrics-that-actually-work-in-cve-impact-prediction-d4bafd9cec1b)<br>    - [GitHub](https://github.com/keerthanap8898/Accuracy-is-Not-Enough-in-Cybersecurity/tree/main)<br>```</pre><h4>IEEE Style :</h4><pre>```text/md<br>    K. Purushotham, “Accuracy Is Not Enough: Confusion Matrix Metrics That Actually Work in CVE Impact Prediction,” Substack, Sep. 2025.  DOI 10.5281/zenodo.17438182 - ( https://doi.org/10.5281/zenodo.17438183 ) .<br>    Also available at:<br>    - [Substack](https://keerthanapurushotham.substack.com/p/accuracy-is-not-enough-confusion)<br>    - [Medium](https://medium.com/@keerthanapurushotham/accuracy-is-not-enough-confusion-matrix-metrics-that-actually-work-in-cve-impact-prediction-d4bafd9cec1b)<br>    - [GitHub](https://github.com/keerthanap8898/Accuracy-is-Not-Enough-in-Cybersecurity/tree/main)<br>```</pre><h4>BibTex Entry :</h4><pre>```bibtex<br>    @misc{purushotham2025accuracy,<br>    author       = {Keerthana Purushotham},<br>    title        = {Accuracy Is Not Enough: Confusion Matrix Metrics That Actually Work in CVE Impact Prediction},<br>    howpublished = {Zenodo},<br>    year         = {2025},<br>    month        = {September},<br>    doi          = {10.5281/zenodo.17438182}<br>    url          = {https://keerthanapurushotham.substack.com/p/accuracy-is-not-enough-confusion},<br>    note         = {Also available on Substack, Medium and GitHub: <br>                    \url{https://keerthanapurushotham.substack.com/p/accuracy-is-not-enough-confusion},<br>                    \url{https://medium.com/@keerthanapurushotham/accuracy-is-not-enough-confusion-matrix-metrics-that-actually-work-in-cve-impact-prediction-d4bafd9cec1b}, <br>                    \url{https://github.com/keerthanap8898/Accuracy-is-Not-Enough-in-Cybersecurity}}<br>                   }<br>```</pre><h4>MODS xml :</h4><pre><br>    &lt;mods xmlns=”http://www.loc.gov/mods/v3” version=”3.7”&gt;<br>      &lt;titleInfo&gt;<br>        &lt;title&gt;Accuracy Is Not Enough&lt;/title&gt;<br>        &lt;subTitle&gt;Confusion Matrix Metrics That Actually Work in CVE Impact Prediction&lt;/subTitle&gt;<br>      &lt;/titleInfo&gt;<br>    <br>      &lt;name type=”personal”&gt;<br>        &lt;namePart&gt;Purushotham, Keerthana&lt;/namePart&gt;<br>        &lt;role&gt;<br>          &lt;roleTerm type=”code” authority=”marcrelator”&gt;aut&lt;/roleTerm&gt;<br>        &lt;/role&gt;<br>      &lt;/name&gt;<br>    <br>      &lt;typeOfResource&gt;text&lt;/typeOfResource&gt;<br>      &lt;genre&gt;Research article&lt;/genre&gt;<br>    <br>      &lt;originInfo&gt;<br>        &lt;publisher&gt;Zenodo&lt;/publisher&gt;<br>        &lt;place&gt;<br>          &lt;placeTerm type=”text”&gt;Online&lt;/placeTerm&gt;<br>        &lt;/place&gt;<br>        &lt;dateIssued encoding=”w3cdtf” keyDate=”yes”&gt;2025-09-19&lt;/dateIssued&gt;<br>      &lt;/originInfo&gt;<br>    <br>      &lt;language&gt;<br>        &lt;languageTerm type=”code” authority=”iso639-2b”&gt;eng&lt;/languageTerm&gt;<br>      &lt;/language&gt;<br>    <br>      &lt;identifier type=”doi”&gt;10.5281/zenodo.17438182&lt;/identifier&gt;<br>      &lt;identifier type=”uri”&gt;https://doi.org/10.5281/zenodo.17438183&lt;/identifier&gt;<br>    <br>      &lt;location&gt;<br>        &lt;url usage=”primary display”&gt;https://doi.org/10.5281/zenodo.17438183&lt;/url&gt;<br>      &lt;/location&gt;<br>    <br>      &lt;relatedItem type=”otherVersion” displayLabel=”Also available at”&gt;<br>        &lt;location&gt;<br>          &lt;url&gt;https://keerthanapurushotham.substack.com/p/accuracy-is-not-enough-confusion&lt;/url&gt;<br>        &lt;/location&gt;<br>      &lt;/relatedItem&gt;<br>    <br>      &lt;relatedItem type=”otherVersion”&gt;<br>        &lt;location&gt;<br>          &lt;url&gt;https://medium.com/@keerthanapurushotham/accuracy-is-not-enough-confusion-matrix-metrics-that-actually-work-in-cve-impact-prediction-d4bafd9cec1b&lt;/url&gt;<br>        &lt;/location&gt;<br>      &lt;/relatedItem&gt;<br>    <br>      &lt;relatedItem type=”otherVersion”&gt;<br>        &lt;location&gt;<br>          &lt;url&gt;https://github.com/keerthanap8898/Accuracy-is-Not-Enough-in-Cybersecurity/tree/main&lt;/url&gt;<br>        &lt;/location&gt;<br>      &lt;/relatedItem&gt;<br>    <br>      &lt;note type=”statement of responsibility”&gt;By Keerthana Purushotham.&lt;/note&gt;<br>    <br>      &lt;recordInfo&gt;<br>        &lt;recordContentSource&gt;Self-deposited via Zenodo&lt;/recordContentSource&gt;<br>        &lt;recordCreationDate encoding=”w3cdtf”&gt;2025-09-19&lt;/recordCreationDate&gt;<br>        &lt;languageOfCataloging&gt;<br>          &lt;languageTerm type=”code” authority=”iso639-2b”&gt;eng&lt;/languageTerm&gt;<br>        &lt;/languageOfCataloging&gt;<br>      &lt;/recordInfo&gt;<br>    &lt;/mods&gt;<br></pre><h4><strong>Mastodon</strong></h4><pre><br>&lt;meta name=&quot;fediverse:creator&quot; content=&quot;@keepur@infosec.exchange&quot;&gt;    <br>&lt;a rel=&quot;me&quot; href=&quot;https://infosec.exchange/@keepur&quot;&gt;Mastodon&lt;/a&gt;<br></pre><h4>References</h4><ul><li><a href="https://encord.com/glossary/confusion-matrix/">What is a Confusion Matrix? | Machine Learning Glossary | Encord | Encord</a></li><li><a href="https://encord.com/blog/f1-score-in-machine-learning/">F1 Score in Machine Learning</a></li><li><a href="https://en.wikipedia.org/wiki/Jaccard_index">Jaccard index - Wikipedia</a></li><li><a href="https://en.wikipedia.org/wiki/Approximate_string_matching">Approximate string matching - Wikipedia</a></li><li><a href="https://en.wikipedia.org/wiki/Fuzzy_logic">Fuzzy logic - Wikipedia</a></li></ul><p>DamerauLevenshtein can be extended from char-strings to set-lvl comparisons for similarity match across full lists instead of tokens only.</p><p><a href="https://en.wikipedia.org/wiki/Damerau%E2%80%93Levenshtein_distance#Applications">Damerau-Levenshtein distance - Wikipedia</a></p><ul><li>Follow me on LinkedIn to keep up with my work more closely.<br><a href="https://www.linkedin.com/in/keerthanapurushotham/">https://www.linkedin.com/in/keerthanapurushotham/</a></li><li>Email: <a href="/service/mailto:keep.consult@proton.me">keep.consult@proton.me</a></li><li><a href="https://github.com/keerthanap8898/Accuracy-is-Not-Enough-in-Cybersecurity">https://github.com/keerthanap8898/Accuracy-is-Not-Enough-in-Cybersecurity</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d4bafd9cec1b" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>