Esempio n. 1
0
        public override void Run()
        {
            // Read files
            IEnumerable <Run> runs = AbstractCommand.ReadInputFile(this._inputPath);
            IEnumerable <RelevanceEstimate> judged = new RelevanceEstimate[] { };

            if (this._judgedPath != null)
            {
                judged = AbstractCommand.ReadKnownJudgments(this._judgedPath);
            }
            // Initialize wrapped estimator
            this._estimator.Initialize(runs, judged);
            // Compile list of all query-doc pairs
            Dictionary <string, HashSet <string> > querydocs = AbstractCommand.ToQueryDocuments(runs);
            // Estimate relevance of all query-doc pairs
            List <RelevanceEstimate> estimates = new List <RelevanceEstimate>();

            foreach (var qd in querydocs)
            {
                foreach (var doc in qd.Value)
                {
                    estimates.Add(this._estimator.Estimate(qd.Key, doc));
                }
            }
            // Output estimates
            TabSeparated io = new TabSeparated(this._decimalDigits);

            ((IWriter <RelevanceEstimate>)io).Write(Console.Out, estimates);
        }
Esempio n. 2
0
        internal static IEnumerable <Metadata> ReadMetadata(string file)
        {
            IEnumerable <Metadata> metadata;

            try {
                IReader <Metadata> reader = new TabSeparated();
                using (StreamReader sr = new StreamReader(File.OpenRead(file))) {
                    metadata = reader.Read(sr);
                }
            } catch (Exception ex) {
                throw new FormatException("Error reading metadata file: " + ex.Message, ex);
            }
            return(metadata);
        }
Esempio n. 3
0
        internal static IEnumerable <RelevanceEstimate> ReadEstimatedJudgments(string file)
        {
            IEnumerable <RelevanceEstimate> estimates = null;

            try {
                IReader <RelevanceEstimate> runReader = new TabSeparated();
                using (StreamReader sr = new StreamReader(File.OpenRead(file))) {
                    estimates = runReader.Read(sr);
                }
            } catch (Exception ex) {
                throw new FormatException("Error reading estimated judgments file: " + ex.Message, ex);
            }
            return(estimates);
        }
Esempio n. 4
0
        // Read files
        internal static IEnumerable <Run> ReadInputFile(string file)
        {
            IEnumerable <Run> runs = null;

            try {
                IReader <Run> runReader = new TabSeparated();
                using (StreamReader sr = new StreamReader(File.OpenRead(file))) {
                    runs = runReader.Read(sr);
                }
            } catch (Exception ex) {
                throw new FormatException("Error reading input file: " + ex.Message, ex);
            }
            return(runs);
        }
Esempio n. 5
0
        public override void Run()
        {
            // Read files
            IEnumerable <Run> runs = AbstractCommand.ReadInputFile(this._inputPath);
            IEnumerable <RelevanceEstimate> judged = new RelevanceEstimate[] { };

            if (this._judgedPath != null)
            {
                judged = AbstractCommand.ReadKnownJudgments(this._judgedPath);
            }
            IEnumerable <RelevanceEstimate> estimates = AbstractCommand.ReadEstimatedJudgments(this._estimatedPath);
            // Instantiate estimate store and measure
            RelevanceEstimateStore store = new RelevanceEstimateStore(estimates);

            store.Update(judged);
            IMeasure measure = new CG(100); //TODO: max relevance

            // Re-structure runs for efficient access
            Dictionary <string, Dictionary <string, Run> > sqRuns = AbstractCommand.ToSystemQueryRuns(runs);

            // Estimate per-query absolute effectiveness
            Dictionary <string, Dictionary <string, AbsoluteEffectivenessEstimate> > sqAbss =
                EvaluateCommand.GetSystemQueryAbsolutes(sqRuns, measure, store, this._confEstimator);
            // Average and sort
            List <AbsoluteEffectivenessEstimate> absSorted = EvaluateCommand.GetSortedMeanAbsolutes(sqAbss, this._confEstimator);

            // Estimate per-query relative effectiveness
            Dictionary <string, Dictionary <string, Dictionary <string, RelativeEffectivenessEstimate> > > ssqRels =
                EvaluateCommand.GetSystemSystemQueryRelatives(sqRuns, measure, store, this._confEstimator);
            // Average (already sorted)
            List <RelativeEffectivenessEstimate> relSorted = EvaluateCommand.GetSortedMeanRelatives(ssqRels, this._confEstimator);

            // Output estimates
            TabSeparated io = new TabSeparated(this._decimalDigits);

            ((IWriter <AbsoluteEffectivenessEstimate>)io).Write(Console.Out, absSorted);
            ((IWriter <RelativeEffectivenessEstimate>)io).Write(Console.Out, relSorted);
        }
Esempio n. 6
0
 public override void Run()
 {
     // Read files
     IEnumerable<Run> runs = AbstractCommand.ReadInputFile(this._inputPath);
     IEnumerable<RelevanceEstimate> judged = new RelevanceEstimate[] { };
     if (this._judgedPath != null) {
         judged = AbstractCommand.ReadKnownJudgments(this._judgedPath);
     }
     // Initialize wrapped estimator
     this._estimator.Initialize(runs, judged);
     // Compile list of all query-doc pairs
     Dictionary<string, HashSet<string>> querydocs = AbstractCommand.ToQueryDocuments(runs);
     // Estimate relevance of all query-doc pairs
     List<RelevanceEstimate> estimates = new List<RelevanceEstimate>();
     foreach (var qd in querydocs) {
         foreach (var doc in qd.Value) {
             estimates.Add(this._estimator.Estimate(qd.Key, doc));
         }
     }
     // Output estimates
     TabSeparated io = new TabSeparated(this._decimalDigits);
     ((IWriter<RelevanceEstimate>)io).Write(Console.Out, estimates);
 }
Esempio n. 7
0
        public override void Run()
        {
            // Read files
            IEnumerable<Run> runs = AbstractCommand.ReadInputFile(this._inputPath);
            IEnumerable<RelevanceEstimate> judged = new RelevanceEstimate[] { };
            if (this._judgedPath != null) {
                judged = AbstractCommand.ReadKnownJudgments(this._judgedPath);
            }
            IEnumerable<RelevanceEstimate> estimates = AbstractCommand.ReadEstimatedJudgments(this._estimatedPath);
            // Instantiate estimate store and measure
            RelevanceEstimateStore store = new RelevanceEstimateStore(estimates);
            store.Update(judged);
            IMeasure measure = new CG(100); //TODO: max relevance

            // Re-structure runs for efficient access
            Dictionary<string, Dictionary<string, Run>> sqRuns = AbstractCommand.ToSystemQueryRuns(runs);

            // Estimate per-query absolute effectiveness
            Dictionary<string, Dictionary<string, AbsoluteEffectivenessEstimate>> sqAbss =
                EvaluateCommand.GetSystemQueryAbsolutes(sqRuns, measure, store, this._confEstimator);
            // Average and sort
            List<AbsoluteEffectivenessEstimate> absSorted = EvaluateCommand.GetSortedMeanAbsolutes(sqAbss, this._confEstimator);

            // Estimate per-query relative effectiveness
            Dictionary<string, Dictionary<string, Dictionary<string, RelativeEffectivenessEstimate>>> ssqRels =
                EvaluateCommand.GetSystemSystemQueryRelatives(sqRuns, measure, store, this._confEstimator);
            // Average (already sorted)
            List<RelativeEffectivenessEstimate> relSorted = EvaluateCommand.GetSortedMeanRelatives(ssqRels, this._confEstimator);

            // Output estimates
            TabSeparated io = new TabSeparated(this._decimalDigits);
            ((IWriter<AbsoluteEffectivenessEstimate>)io).Write(Console.Out, absSorted);
            ((IWriter<RelativeEffectivenessEstimate>)io).Write(Console.Out, relSorted);
        }
Esempio n. 8
0
 internal static IEnumerable<Metadata> ReadMetadata(string file)
 {
     IEnumerable<Metadata> metadata;
     try {
         IReader<Metadata> reader = new TabSeparated();
         using (StreamReader sr = new StreamReader(File.OpenRead(file))) {
             metadata = reader.Read(sr);
         }
     } catch (Exception ex) {
         throw new FormatException("Error reading metadata file: " + ex.Message, ex);
     }
     return metadata;
 }
Esempio n. 9
0
 internal static IEnumerable<RelevanceEstimate> ReadKnownJudgments(string file)
 {
     IEnumerable<RelevanceEstimate> judged = null;
     try {
         IReader<RelevanceEstimate> runReader = new TabSeparated();
         using (StreamReader sr = new StreamReader(File.OpenRead(file))) {
             judged = runReader.Read(sr);
         }
     } catch (Exception ex) {
         throw new FormatException("Error reading known judgments file: " + ex.Message, ex);
     }
     return judged;
 }
Esempio n. 10
0
 // Read files
 internal static IEnumerable<Run> ReadInputFile(string file)
 {
     IEnumerable<Run> runs = null;
     try {
         IReader<Run> runReader = new TabSeparated();
         using (StreamReader sr = new StreamReader(File.OpenRead(file))) {
             runs = runReader.Read(sr);
         }
     } catch (Exception ex) {
         throw new FormatException("Error reading input file: " + ex.Message, ex);
     }
     return runs;
 }
Esempio n. 11
0
        public override void Run()
        {
            // Read files
            IEnumerable <Run> runs = AbstractCommand.ReadInputFile(this._inputPath);
            IEnumerable <RelevanceEstimate> judged = new RelevanceEstimate[] { };

            if (this._judgedPath != null)
            {
                judged = AbstractCommand.ReadKnownJudgments(this._judgedPath);
            }
            IEnumerable <RelevanceEstimate> estimates = AbstractCommand.ReadEstimatedJudgments(this._estimatedPath);
            // Instantiate estimate store and measure
            RelevanceEstimateStore store = new RelevanceEstimateStore(estimates);

            store.Update(judged);
            IMeasure measure = new CG(100); //TODO: max relevance

            // Compile list of all query-doc-sys-rank tuples
            Dictionary <string, Dictionary <string, Dictionary <string, int> > > qdsRanks = AbstractCommand.ToQueryDocumentSystemRanks(runs);
            // Re-structure estimates
            Dictionary <string, Dictionary <string, RelevanceEstimate> > qdEstimates = new Dictionary <string, Dictionary <string, RelevanceEstimate> >();

            foreach (var est in estimates)
            {
                Dictionary <string, RelevanceEstimate> dEstimates = null;
                if (!qdEstimates.TryGetValue(est.Query, out dEstimates))
                {
                    dEstimates = new Dictionary <string, RelevanceEstimate>();
                    qdEstimates.Add(est.Query, dEstimates);
                }
                dEstimates.Add(est.Document, est);
            }
            // Remove judged query-docs
            foreach (var j in judged)
            {
                Dictionary <string, RelevanceEstimate> dEstimates = null;
                if (qdEstimates.TryGetValue(j.Query, out dEstimates))
                {
                    dEstimates.Remove(j.Document);
                    if (dEstimates.Count == 0)
                    {
                        qdEstimates.Remove(j.Query);
                    }
                }
            }

            bool needsNext = false;
            // Re-structure runs for efficient access
            Dictionary <string, Dictionary <string, Run> > sqRuns = AbstractCommand.ToSystemQueryRuns(runs);

            if (this._target == EvaluationTargets.Relative)
            {
                // Estimate per-query relative effectiveness
                Dictionary <string, Dictionary <string, Dictionary <string, RelativeEffectivenessEstimate> > > ssqRels =
                    EvaluateCommand.GetSystemSystemQueryRelatives(sqRuns, measure, store, this._confEstimator);
                // Average (already sorted)
                List <RelativeEffectivenessEstimate> relSorted = EvaluateCommand.GetSortedMeanRelatives(ssqRels, this._confEstimator);

                if (relSorted.Average(r => r.Confidence) < this._confidence)
                {
                    needsNext = true;
                    measure.ComputeQueryDocumentWeights(qdEstimates, qdsRanks, ssqRels);
                }
            }
            else
            {
                // Estimate per-query absolute effectiveness
                Dictionary <string, Dictionary <string, AbsoluteEffectivenessEstimate> > sqAbss =
                    EvaluateCommand.GetSystemQueryAbsolutes(sqRuns, measure, store, this._confEstimator);
                // Average and sort
                List <AbsoluteEffectivenessEstimate> absSorted = EvaluateCommand.GetSortedMeanAbsolutes(sqAbss, this._confEstimator);

                if (absSorted.Average(a => a.Confidence) < this._confidence)
                {
                    needsNext = true;
                    measure.ComputeQueryDocumentWeights(qdEstimates, qdsRanks, sqAbss);
                }
            }

            if (needsNext)
            {
                var          batches = NextCommand.GetBatches(qdEstimates, this._batchNum, this._batchSize);
                TabSeparated io      = new TabSeparated(this._decimalDigits);
                ((IWriter <List <RelevanceEstimate> >)io).Write(Console.Out, batches.Take(this._batchNum));
            }
        }
Esempio n. 12
0
        public override void Run()
        {
            // Read files
            IEnumerable<Run> runs = AbstractCommand.ReadInputFile(this._inputPath);
            IEnumerable<RelevanceEstimate> judged = new RelevanceEstimate[] { };
            if (this._judgedPath != null) {
                judged = AbstractCommand.ReadKnownJudgments(this._judgedPath);
            }
            IEnumerable<RelevanceEstimate> estimates = AbstractCommand.ReadEstimatedJudgments(this._estimatedPath);
            // Instantiate estimate store and measure
            RelevanceEstimateStore store = new RelevanceEstimateStore(estimates);
            store.Update(judged);
            IMeasure measure = new CG(100); //TODO: max relevance

            // Compile list of all query-doc-sys-rank tuples
            Dictionary<string, Dictionary<string, Dictionary<string, int>>> qdsRanks = AbstractCommand.ToQueryDocumentSystemRanks(runs);
            // Re-structure estimates
            Dictionary<string, Dictionary<string, RelevanceEstimate>> qdEstimates = new Dictionary<string, Dictionary<string, RelevanceEstimate>>();
            foreach (var est in estimates) {
                Dictionary<string, RelevanceEstimate> dEstimates = null;
                if (!qdEstimates.TryGetValue(est.Query, out dEstimates)) {
                    dEstimates = new Dictionary<string, RelevanceEstimate>();
                    qdEstimates.Add(est.Query, dEstimates);
                }
                dEstimates.Add(est.Document, est);
            }
            // Remove judged query-docs
            foreach (var j in judged) {
                Dictionary<string, RelevanceEstimate> dEstimates = null;
                if (qdEstimates.TryGetValue(j.Query, out dEstimates)) {
                    dEstimates.Remove(j.Document);
                    if (dEstimates.Count == 0) {
                        qdEstimates.Remove(j.Query);
                    }
                }
            }

            bool needsNext = false;
            // Re-structure runs for efficient access
            Dictionary<string, Dictionary<string, Run>> sqRuns = AbstractCommand.ToSystemQueryRuns(runs);

            if (this._target == EvaluationTargets.Relative) {
                // Estimate per-query relative effectiveness
                Dictionary<string, Dictionary<string, Dictionary<string, RelativeEffectivenessEstimate>>> ssqRels =
                    EvaluateCommand.GetSystemSystemQueryRelatives(sqRuns, measure, store, this._confEstimator);
                // Average (already sorted)
                List<RelativeEffectivenessEstimate> relSorted = EvaluateCommand.GetSortedMeanRelatives(ssqRels, this._confEstimator);

                if (relSorted.Average(r => r.Confidence) < this._confidence) {
                    needsNext = true;
                    measure.ComputeQueryDocumentWeights(qdEstimates, qdsRanks, ssqRels);
                }
            } else {
                // Estimate per-query absolute effectiveness
                Dictionary<string, Dictionary<string, AbsoluteEffectivenessEstimate>> sqAbss =
                    EvaluateCommand.GetSystemQueryAbsolutes(sqRuns, measure, store, this._confEstimator);
                // Average and sort
                List<AbsoluteEffectivenessEstimate> absSorted = EvaluateCommand.GetSortedMeanAbsolutes(sqAbss, this._confEstimator);

                if (absSorted.Average(a => a.Confidence) < this._confidence) {
                    needsNext = true;
                    measure.ComputeQueryDocumentWeights(qdEstimates, qdsRanks, sqAbss);
                }
            }

            if (needsNext) {
                var batches = NextCommand.GetBatches(qdEstimates, this._batchNum, this._batchSize);
                TabSeparated io = new TabSeparated(this._decimalDigits);
                ((IWriter<List<RelevanceEstimate>>)io).Write(Console.Out, batches.Take(this._batchNum));
            }
        }