Exemple #1
0
        public async Task Process(IConnectionContext target, TrainMessage request, CancellationToken token)
        {
            if (target == null)
            {
                throw new ArgumentNullException(nameof(target));
            }
            if (request == null)
            {
                throw new ArgumentNullException(nameof(request));
            }

            ISentimentDataHolder loader = default;
            var completed = new CompletedMessage();

            try
            {
                if (!string.IsNullOrEmpty(request.Domain))
                {
                    logger.LogInformation("Using Domain dictionary [{0}]", request.Domain);
                    loader = lexiconLoader.GetLexicon(request.Domain);
                }

                var modelLocation = storage.GetLocation(target.Connection.User, request.Name, ServiceConstants.Model);

                using (var scope = provider.CreateScope())
                {
                    var container = scope.ServiceProvider.GetService <ISessionContainer>();
                    container.Context.NGram = 3;
                    var client    = container.GetTraining(modelLocation);
                    var converter = scope.ServiceProvider.GetService <IDocumentConverter>();
                    client.Pipeline.ResetMonitor();
                    if (loader != null)
                    {
                        client.Lexicon = loader;
                    }

                    var positive = storage.Load(target.Connection.User, request.Name, true)
                                   .Take(2000);
                    var negative = storage.Load(target.Connection.User, request.Name, false)
                                   .Take(2000);

                    var documents = positive.Concat(negative)
                                    .Select(item => converter.Convert(item, request.CleanText));
                    await client.Train(documents).ConfigureAwait(false);

                    completed.Message = "Training Completed";
                    await target.Write(completed, token).ConfigureAwait(false);
                }
            }
            catch (Exception e)
            {
                completed.Message = e.Message;
                await target.Write(completed, token).ConfigureAwait(false);

                completed.IsError = true;
                throw;
            }
        }
        public async Task Process(IConnectionContext target, SentimentMessage message, CancellationToken token)
        {
            if (message == null)
            {
                throw new ArgumentNullException(nameof(message));
            }

            var request = message.Request;

            if (request?.Documents == null)
            {
                throw new Exception("Nothing to process");
            }

            if (request.Documents.Length > 500)
            {
                throw new Exception("Too many documents. Maximum is 500");
            }

            var completed = new CompletedMessage();

            try
            {
                var monitor = new PerformanceMonitor(request.Documents.Length);

                using (Observable.Interval(TimeSpan.FromSeconds(10))
                       .Subscribe(item => logger.LogInformation(monitor.ToString())))
                {
                    ISentimentDataHolder lexicon = default;

                    if (request.Dictionary != null &&
                        request.Dictionary.Count > 0)
                    {
                        logger.LogInformation("Creating custom dictionary with {0} words", request.Dictionary.Count);

                        lexicon = SentimentDataHolder.Load(request.Dictionary.Select(item =>
                                                                                     new WordSentimentValueData(
                                                                                         item.Key,
                                                                                         new SentimentValueData(item.Value))));
                    }

                    if ((lexicon == null || request.AdjustDomain) &&
                        !string.IsNullOrEmpty(request.Domain))
                    {
                        logger.LogInformation("Using Domain dictionary [{0}]", request.Domain);
                        var previous = lexicon;
                        lexicon = lexiconLoader.GetLexicon(request.Domain);
                        if (previous != null)
                        {
                            lexicon.Merge(previous);
                        }
                    }

                    string modelLocation = null;

                    if (!string.IsNullOrEmpty(request.Model))
                    {
                        logger.LogInformation("Using model path: {0}", request.Model);
                        modelLocation = storage.GetLocation(target.Connection.User, request.Model, ServiceConstants.Model);

                        if (!Directory.Exists(modelLocation))
                        {
                            throw new ApplicationException($"Can't find model {request.Model}");
                        }
                    }

                    using (var scope = provider.CreateScope())
                    {
                        var container = scope.ServiceProvider.GetService <ISessionContainer>();
                        container.Context.NGram             = 3;
                        container.Context.ExtractAttributes = request.Emotions;

                        var client    = container.GetTesting(modelLocation);
                        var converter = scope.ServiceProvider.GetService <IDocumentConverter>();
                        client.Init();
                        client.Pipeline.ResetMonitor();

                        if (lexicon != null)
                        {
                            client.Lexicon = lexicon;
                        }

                        await client.Process(request.Documents.Select(item => converter.Convert(item, request.CleanText))
                                             .ToObservable())
                        .Select(item =>
                        {
                            monitor.Increment();
                            return(item);
                        })
                        .Buffer(TimeSpan.FromSeconds(5), 10, scheduler)
                        .Select(async item =>
                        {
                            var result = new ResultMessage <Document> {
                                Data = item.Select(x => x.Processed).ToArray()
                            };
                            await target.Write(result, token).ConfigureAwait(false);
                            return(Unit.Default);
                        })
                        .Merge();
                    }

                    logger.LogInformation("Completed with final performance: {0}", monitor);
                    completed.Message = "Testing Completed";
                    await target.Write(completed, token).ConfigureAwait(false);
                }
            }
            catch (Exception e)
            {
                completed.Message = e.Message;
                await target.Write(completed, token).ConfigureAwait(false);

                completed.IsError = true;
                throw;
            }
        }