Binary trees? Automatically identifying the links between born-digital records

Figures & data

Table 1. Checksum algorithms and string-length comparisons.

Algorithm	String length	Author
MD5	32	Ronald Rivest, 1992
SHA1	40	US National Security Agency, 1993
SHA256	64	US National Security Agency, 2001

Table 2. Checksum digests are lexically different to one-another, even when differences between two files are small.

Example	Checksum
File A: with zero changes	8c69dc0668c4c73092a7042df45e756adb170742
File B: with the first byte removed	6b75b8f235c148efd1b03d9c113664895b5aa7cd

Table 3. In contrast to cryptographic hashes, fuzzy hashes enables granular comparison, even when two digital files are not identical.

Example	Fuzzy hash
File A: with zero changes	1536:tLQy16aYRCWYTESSg3yDuBCwclnHpQ/B4kCK7ZBEY0t5vykp6CYP:q1aYpYTESSgM2CwQGt9ZBB1U6hP
File C: with the first 250 bytes removed	1536:CLQy16aYRCWYTESSg3yDuBCwclnHpQ/B4kCK7ZBEY0t5vykp6CYP:B1aYpYTESSgM2CwQGt9ZBB1U6hP

Table

00	0F	00	05	10	34	00	00	00	00	00	04	10	08	00	00	00	04
00	00	00	00	00	33	01	00	00	BA	0F	1C	00	00	00	6D	00	2D
00	74	00	6F	00	76	00	2D	00	73	00	68	00	65	00	6D	00	2E
00	77	00	6D	00	76	00	0F	00	07	10	38

Figure 1. Example entity highlighting using Stanford’s Named Entity Recognizer.