Freelancer.com, the world\u2019s largest freelancing and\ncrowdsourcing marketplace by number of users and jobs posted, announced the\nwinners of the Risky Space Business Challenge run in collaboration with\ngovernment-focused consultancy LMI, on behalf of NASA.<\/p>\n\n\n\n
Announced on 22 October 2021, the challenge gave\nparticipants 15 weeks to develop an artificial intelligence (AI) and machine\nlearning (ML) algorithm that can collect and analyze data from past projects in\nany organization to allow teams to predict, prevent, and prepare for different\nrisks or pitfalls involved in the project at hand.<\/p>\n\n\n\n
The challenge received 27 submissions offering NASA new and\ninnovative solutions to support existing risk prediction tools. Of those\nentries, six projects came out on top and were awarded a share of US$50,000 as\nevaluated by LMI\u2019s subject matter experts and a panel of NASA experts.<\/p>\n\n\n\n
Winners – Top Solutions<\/strong><\/p>\n\n\n\n First Place<\/strong> – winning US$20,000: Christopher Milo (Team Lead) from Ashburn, Virginia, United States, supported by team members Zach Pryor, Benjamin Walzer, Daniel Mask, Jacob Walzer, Sean Mellott.<\/p>\n\n\n\n The winning solution employed algorithms programmed in\nPython that leveraged open-source ML and Natural Language Processing (NLP)\nmodels to train text classifiers on historical NASA documents to recognize\nlanguage patterns relating to risk. <\/p>\n\n\n\n Second Place<\/strong> – winning US$10,000: Rich\u00c3\u00a1rd \u00c3\u0081d\u00c3\u00a1m V\u00e9csey Dr. (Team Lead) from Budapest, Hungary, supported by his team member Axel Orsz\u00c3\u00a1g-Krisz Dr.<\/p>\n\n\n\n The solution predicts risks in two different ways from\nproject-related text documents: 5×5 risk severity-likelihood matrix and a list\nof risk categories. It contains three different neural network models – one\npredicts the chance of the occurrences of each unique risk category in the\ntext, the second predicts severity and likelihood values for each risk, the\nthird recognizes whether the text contains any risk or not.<\/p>\n\n\n\n Third Place<\/strong> – winning US$5,000: Thomas Ilin from Thornhaugh, United Kingdom.<\/p>\n\n\n\n The proposal focuses on a solution architecture that can\nmake a truly ‘Game Changing’ difference to the way project risks are identified\nand predicted by NASA, across all of its directorates and projects.<\/p>\n\n\n\n \u201cThe winners of NASA’s Risky Space Business Challenge\ndemonstrate how crowdsourcing can provide organizations with exciting and novel\napproaches to complex problems. Each and every entry we had submitted provided\nan intriguing solution, but the six winners produced some truly outstanding\nwork,\u201d said Freelancer.com\u2019s Chief Executive Matt Barrie.<\/p>\n\n\n\n Winners – Innovation Solutions:<\/strong><\/p>\n\n\n\n Risk Prediction<\/strong> – winning US$5,000: Alexander Poplavsky from Krakow, Poland.<\/p>\n\n\n\n The proposed solution comprises several AI text processing\nmodels and related algorithms to estimate the unknown project risk based on the\nhistorical Lessons Learned data and a target Project Plan information. The risk\nis estimated following the standard NASA classification: cost, schedule,\ntechnical, and programmatic affinities with green, yellow and red levels.<\/p>\n\n\n\n Data Extraction<\/strong> – winning US$5,000: Petra Galuscakova from Drazkovce, Slovakia.<\/p>\n\n\n\n The solution reformulates the problem of predicting the\nproject risks to the problem of searching similar earlier NASA projects. It\nfocuses on creating a searchable index of the materials, such as lessons\nlearned, presentations and proposals, used in the earlier projects.<\/p>\n\n\n\n Data Formatting<\/strong> – winning US$5,000: Dean Koucoulas (team lead) from Etobicoke, Canada, supported by team member Snezana Kirova.<\/p>\n\n\n\n The solution implemented techniques in Natural Language\nProcessing in Python to compare the information within project documents with a\nrisk-based lexicon\/dictionary. The information is then ultimately categorized\naccording to what Condition, Departure, Asset, and Consequence best match the\nrisk details, drawing upon classification schemes inherent in the NASA\nContinuous Risk Management Process.<\/p>\n\n\n\n \u201cThe challenge proved the value of crowdsourcing through the\nidentification of numerous unique approaches that combine text extraction and\nanalytics with novel risk prediction methods and algorithms,\u201d said Brian Tonge,\nPrincipal, Advanced Analytics & AI at LMI. \u201cThe collective solution\nportfolio supplies NASA\u2019s engineers, program managers and analysts with an\narray of options that can be further refined and deployed to achieve an\nadvanced risk prediction capability that can transform program management and\ndecision-making.\u201d <\/p>\n\n\n\n