SUSTAINABLE AQUACULTURE: BENCHMARKING STANDARDS, IMPROVING CERTIFICATION

Currently there are a plethora of well-established certification schemes with various focal points. These include management/food quality, organic production and national coverage.  There are also upcoming standards that have a regional international scope. Given the large number of standards, benchmarking using criteria that lead to sustainable aquaculture is necessary. Benchmarking can either be quantitative or qualitative in nature but results so far have been based on subjective classifications. Here we present an objective benchmarking that relies on machine learning using natural language processing of text content from standards. Four shrimp standards namely: ASC, BAP. Naturland and SEASAIP were subjected to text mining and four objective topic areas were mined using latent dirichlet allocation (LDA). Naming of the topics was done through mind mapping procedure that involved an international audience.

The names derived include: Pre-requisites, Analytics, Management and Regulation. The four standards under examination have different probabilities for each topic with ASC having 82% inclination towards pre-requisites, BAP leans towards management at 88.2%, Naturland is more interested in regulation (95.3%) while SEASAIP (88.7%) tends to be in favour of analytics for better decision making.

Furthermore, audits are meant to give objective information through independent evaluation that proves the existence of a quality system compliant with a given standard. Audits are usually a snapshot view of compliance that is done within a day hence preparations can be made to effectively hide non-compliance. We also present here a suite of new innovations in the social sphere that are aimed at continuous 'real-time' data acquisition for monitoring labour.