{"id":21529,"date":"2022-12-07T09:55:13","date_gmt":"2022-12-07T09:55:13","guid":{"rendered":"https:\/\/www.imprima.com\/?p=21529"},"modified":"2023-02-28T12:17:57","modified_gmt":"2023-02-28T12:17:57","slug":"automated-pii-redaction","status":"publish","type":"post","link":"https:\/\/www.imprima.com\/blog\/automated-pii-redaction","title":{"rendered":"Automated PII Redaction"},"content":{"rendered":"\n

Imprima Smart Redaction achieves accuracy (recall) of around 95%<\/strong><\/p>\n\n\n\n

Sellers, or their advisors, are often confronted with the requirement to redact Personally Identifiable Information – an error-prone and laborious task. Outsourcing is expensive and may cause delays before your deal can go live. And why spend money on it, when it can be done automatically, in your own VDR?<\/p>\n\n\n\n

Your concern may be its reliability? Well, no need to be concerned. We have shown that it is very reliable. In fact, Imprima\u2019s technology will find in most cases at least 90% of the terms to be redacted. Not enough? Well, consider this: a 2011 scientific study by Grossmann and Cormack showed that manual reviews often deliver low\u202freliability, ranging between 50% and 85% (\u201cRecall\u201d). We see that phenomena in our VDRs too. When we test our technology on documents that have already been manually redacted, we see that it finds a lot of items (names, addresses etc.) that humans missed. <\/p>\n\n\n\n

<\/div>\n\n\n\n
\"\"
Test results of Imprima Automatic Redaction on a combination of English, German, Italian and Dutch documents, randomly picked from Imprima VDRs, where the objective was to automatically redact: person name, address etc., see table. Imprima\u2019s machine-learning algorithm (a native tool in the Imprima VDR) was applied. As can be seen from the results in the table, a very high recall is achieved for all entities, in all languages, where overall recall is higher than 90% for all items redacted. So, what does this mean? Recall is a measure of\u202freliability, in particular how many of the items to be redacted are actually found. Not shown in the table, but this was also achieved with very high precision (a measure for how few words were redacted that did not need to be redacted): only a very tiny fraction of all words in the document were \u201cover redacted\u201d. If you would like more information on this experiment, please contact Imprima<\/a>.<\/figcaption><\/figure>\n\n\n\n
<\/div>\n\n\n