The Role of Identity Resolution in Criminal Investigations

Homeland Security, Identity Resolution, Name Matching

entity matching

Law enforcement agencies investigate many forms of criminal activity, some by single perpetrators, others by organized networks of bad actors such as drug traffickers, gang members, smugglers, or human traffickers. Criminal investigations are typically very labor intensive and time critical to protect our communities from further harm.

One crucial but labor intensive and time consuming part of criminal investigations is the process of checking suspects against multiple databases such as arrest records, case notes, watch lists, and telephone records. There are two ways in which this database analysis is important. First, it can help identify potential suspects or provide additional information about an existing suspect. Second, especially in an organized crime scenario, it can reveal previously unknown connections or relationships within a network.

However, record matching can be like finding a needle in a haystack. The amount of data to search through can be staggering and it’s often not as simple as looking for an exact name match since records across databases may contain misspellings, partial names, nicknames, abbreviations, word ordering discrepancies, etc. This data matching problem is compounded when the investigation involves international databases or names from other language scripts (e.g., Cyrillic, Arabic), which may have been transliterated into Latin script in multiple ways and exhibit ethnicity-specific phenomena. What’s more, probable database matches need to be based on similarities other than names. Some match in attributes such as place of birth, address, associates, relatives, and aliases is often needed to ensure more accurate matches.

Identity Resolution Software for Criminal Investigations

The good news is that this complex database analysis can be automated using modern analytical tools. Specifically advanced Entity Matching, also known as Identity Resolution, is critical not only to speed up the record matching process but also to uncover records and connections hidden in large and imperfect data.

For instance, given a search for a name like Michael Stevens, Entity Matching software should be able to return probable matches like Michael Andrew Stevens, Mike Steven, M. Steven, or Steven Michael based not just on name similarity but also other key attributes like aliases, date of birth, place of birth, address, relatives, or associates.

Entity Matching can also reveal previously unknown connections between bad actors. For instance, given a known bad actor, Entity Matching can return records for other bad actors that share some property such as the same or similar address or a common associate, thus helping build a more complete understanding of a criminal network.

NetOwl’s EntityMatcher software performs fuzzy matching. It can match entities with different name spellings, ethnic-specific phenomena (e.g., optional ‘al’), missing components (e.g., middle name), and transliterations. It can also match names across different language scripts such as Cyrillic, Chinese, Arabic, and Farsi.

NetOwl’s Machine Learning-based entity matching product achieves highly accurate identity resolution using probabilistic rules derived from large-scale, real-world, multi-ethnicity name and other attribute data. It performs identity resolution against tens of millions or more entity records in near-real time, with sub-second response times. It is customizable, allowing parameter tuning as well as addition of custom rules and dictionaries.

NetOwl’s EntityMatcher can perform seamlessly through both structured and unstructured data when combined with NetOwl’s Extractor. For more information on NetOwl’s identity resolution software, contact NetOwl today.