Example #1
0
        private static SequenceStats ProcessMinHash(
            DuplicateRates groundTruth,
            Dictionary <string, string> documents,
            int hashCount)
        {
            Console.WriteLine("Min: " + hashCount + " " + DateTime.Now);

            var analyzer = new HashDuplicateAnalyzer <ulong>(new MinDuplicateHasher(hashCount, 5));

            return(Process(groundTruth, documents, analyzer));
        }
Example #2
0
        private static SequenceStats ProcessSimHash(
            DuplicateRates groundTruth,
            Dictionary <string, string> documents,
            int bytesCount)
        {
            Console.WriteLine("Sim: " + bytesCount * 8 + " " + DateTime.Now);

            var analyzer = new HashDuplicateAnalyzer <byte>(new SimDuplicateHasher(bytesCount, 5));

            return(Process(groundTruth, documents, analyzer));
        }