public override void Run()
{
// Read files, initialize store, estimator and measure
IEnumerable<Run> runs = AbstractCommand.ReadInputFile(this._inputPath);
RelevanceEstimateStore store = new RelevanceEstimateStore(AbstractCommand.ReadKnownJudgments(this._judgedPath));
this._estimator.Initialize(runs, new RelevanceEstimate[] { }); // No known judgments at this point
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 runs for efficient access
Dictionary<string, Dictionary<string, Run>> sqRuns = AbstractCommand.ToSystemQueryRuns(runs);
// Estimate relevance of all query-doc pairs
Dictionary<string, Dictionary<string, RelevanceEstimate>> qdEstimates = new Dictionary<string, Dictionary<string, RelevanceEstimate>>();
foreach (var dsRanks in qdsRanks) {
string query = dsRanks.Key;
Dictionary<string, RelevanceEstimate> dEstimates = new Dictionary<string, RelevanceEstimate>();
foreach (var sRanks in dsRanks.Value) {
string doc = sRanks.Key;
dEstimates.Add(doc, this._estimator.Estimate(query,doc)); // Don't estimate yet, will do inside the loop
}
qdEstimates.Add(query, dEstimates);
}
bool needsNext = true;
double confidence = 0.5;
int iteration = 1, judged = 0;
do {
/* Evaluate */
if (this._target == EvaluationTargets.Relative) {
// Estimate per-query relative effectiveness
Dictionary<string, Dictionary<string, Dictionary<string, RelativeEffectivenessEstimate>>> ssqRels =
EvaluateCommand.GetSystemSystemQueryRelatives(sqRuns, measure, this._estimator, this._confEstimator);
// Average (already sorted)
List<RelativeEffectivenessEstimate> relSorted = EvaluateCommand.GetSortedMeanRelatives(ssqRels, this._confEstimator);
confidence = relSorted.Average(r => r.Confidence);
if (confidence < this._confidence) {
measure.ComputeQueryDocumentWeights(qdEstimates, qdsRanks, ssqRels);
} else {
needsNext = false;
}
} else {
// Estimate per-query absolute effectiveness
Dictionary<string, Dictionary<string, AbsoluteEffectivenessEstimate>> sqAbss =
EvaluateCommand.GetSystemQueryAbsolutes(sqRuns, measure, this._estimator, this._confEstimator);
// Average and sort
List<AbsoluteEffectivenessEstimate> absSorted = EvaluateCommand.GetSortedMeanAbsolutes(sqAbss, this._confEstimator);
confidence = absSorted.Average(a => a.Confidence);
if (confidence < this._confidence) {
measure.ComputeQueryDocumentWeights(qdEstimates, qdsRanks, sqAbss);
} else {
needsNext = false;
}
}
Console.WriteLine(iteration + " : Conf=" + confidence + " Judged=" + judged);
if (needsNext) {
/* Next */
var batches = NextCommand.GetBatches(qdEstimates, this._batchNum, this._batchSize);
// "Judge" all batches
foreach (var batch in batches) {
Console.WriteLine(batch[0].Query + " : " + string.Join(" ", batch.Select(d => d.Document)));
foreach (var doc in batch) {
this._estimator.Update(store.Estimate(doc.Query, doc.Document));
judged++;
// Remove from list of pending
var dEstimates = qdEstimates[doc.Query];
dEstimates.Remove(doc.Document);
if (dEstimates.Count == 0) {
qdEstimates.Remove(doc.Query);
}
}
}
/* Re-estimate */
// Re-estimate relevance of pending query-doc pairs
foreach (var dEstimates in qdEstimates) {
string query = dEstimates.Key;
Dictionary<string, RelevanceEstimate> estimates = dEstimates.Value;
foreach (string doc in estimates.Keys.ToArray()) {
estimates[doc] = this._estimator.Estimate(query, doc);
}
}
}
iteration++;
} while (needsNext);
// TODO: output effectiveness estimates
}
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));
}
}