Effective adverse media searches are an important part of risk-based regulatory compliance and enable organizations to quickly identify accurate, actionable information about their customers. With that in mind, customer name matching is significantly complicated when organizations conduct searches across foreign language media sources which use unfamiliar names, naming conventions, non-Western characters, and grammatical conventions.
Multilingual name matching represents a significant risk concern: organizations that miss important news stories involving their customers can face compliance fines and even criminal penalties. However, with significant variation in both the spoken and written language systems of different cultures, adverse media name searches can easily generate vast amounts of data – overloading compliance teams with false positive matches that must be remediated on a case by case basis.
With that in mind, it is crucial that organizations with specific regulatory compliance obligations implement adverse media solutions capable of managing the challenges of multilingual name matching, including the following key considerations:
Language variety and classification: When setting adverse media screening parameters, organizations should consider which countries, and languages, should be included in the scope of their searches. Customers may have a residence in more than one country and so potentially feature in stories from a number of different foreign-language media outlets.
With that in mind, when setting search parameters, organizations should select screening solutions that are able to accurately review unstructured news media data from across different outlets, in the relevant languages. Those solutions should ideally be combined with machine learning-based data classifiers that help determine the significance of news stories and generate more accurate actionable data.
Publication credibility: Screening adverse media across different languages inevitably involves examining stories from multiple publications of varying quality, credibility, and bias. Organizations must not only identify foreign language stories involving their clients but be sensitive to the contexts in which those stories appear.
In practice, this means implementing screening systems that can be trained to capture and remediate the nuances of local languages or take into account journalism quality, cultural influences, and political biases. Equipped with a better understanding of media sources, firms can better process, and assign significance to, the foreign language stories that trigger alerts.
Naming conventions: While organizations may implicitly understand naming conventions in their own jurisdictions, foreign cultures may use significantly different conventions. Western countries use a first name, surname convention when writing names, but that order is reversed in most Asian cultures, which instead use a surname, first name convention. Similarly, many Middle Eastern cultures attach the prefix ‘al’ to denote surnames.
When many foreign language speakers translate their name into English, they approximate the anglicized spelling which may result in lexical variations between users – such as ‘Osama/Usama’. Similarly, certain cultures may use nicknames with a greater level of formality than others, with customers sometimes using those nicknames on official documentation. Accordingly, adverse media solutions must account for spelling approximations and cultural idiosyncrasies by factoring in the variant spellings and conventions.
Non-Latinate characters: European language systems broadly use a Latinate alphabet with a standardized set of 26 characters, but many foreign languages do not. Slavic countries, for example, use the Cyrillic alphabet, China uses the Mandarin alphabet, and many Middle Eastern countries use the Arabic alphabet: each of those alphabets use a unique set of characters, complete with different syntactic rules and naming conventions.
Given the potential diversity of language systems, adverse media solutions must account for name spellings in the Western alphabet and in any relevant foreign alphabets. Practically, this means integrating a multilingual name matching tool with text analytics and transliteration capabilities for the languages relevant to an organization’s search parameters.
Misidentification and duplication: Every culture has naming trends that lead to similarities and homogenization, and that make name misidentification more likely. In the Middle East, for example, the name ‘Mohammed’ has numerous variant spellings, including ‘Mohamed’ and ‘Muhammad’: the proximity of the spellings may increase the potential of a false positive customer name match during an adverse media search.
Similarly, high profile customers may feature in a wide variety of news stories or in a number of very similar stories with varying relevance to compliance risk management. In these contexts, adverse media searches must be able to identify redundant and duplicated stories, and identify duplicate names, in order to reduce noise during the screening process.
Adverse Media Screening Technology
Since multilingual name matching represents a significant compliance burden, it is crucial that firms integrate a suitable software platform to meet their regulatory obligations. In addition to automated speed and efficiency, the software can empower firms to address the specific challenges of multilingual name matching with sophisticated language processing and identity resolutions tools.
Adverse media solutions may also be empowered by machine learning systems that harness customer data to make intuitive decisions about the vast amounts of unstructured adverse media that they collect, using fuzzy logic and algorithmic analysis to disregard false positives and identify true positives, with greater speed and efficiency. Adverse media software can also be updated continuously to reflect a changing compliance risk environment, keeping firms protected as customer profiles change, and screening millions of foreign language news sources in near real time.