private void ExpandThemes()
{
FeatureTrackingManager.Instance.UseFeature(Features.Brainstorm_ExploreLibrary_Document_Themes);
bool added_at_least_one_theme = false;
if (null != pdf_document_node_content.PDFDocument.Library.ExpeditionManager)
{
ExpeditionDataSource eds = pdf_document_node_content.PDFDocument.Library.ExpeditionManager.ExpeditionDataSource;
if (null != eds)
{
if (eds.docs_index.ContainsKey(pdf_document_node_content.PDFDocument.Fingerprint))
{
int doc_id = eds.docs_index[pdf_document_node_content.PDFDocument.Fingerprint];
TopicProbability[] topics = eds.LDAAnalysis.DensityOfTopicsInDocsSorted[doc_id];
for (int t = 0; t < topics.Length && t < 5; ++t)
{
string topic_name = eds.GetDescriptionForTopic(topics[t].topic, false, "\n");
ThemeNodeContent tnc = new ThemeNodeContent(topic_name, pdf_document_node_content.PDFDocument.Library.WebLibraryDetail.Id);
NodeControlAddingByKeyboard.AddChildToNodeControl(node_control, tnc, false);
added_at_least_one_theme = true;
}
}
}
}
if (!added_at_least_one_theme)
{
MessageBoxes.Warn("There were no themes available for this document. Please run Expedition against your library.");
}
}
internal static void AddDocumentsSimilarToDistribution(NodeControl node_control_, WebLibraryDetail web_library_detail, ExpeditionDataSource eds, float[] tags_distribution)
{
WPFDoEvents.AssertThisCodeIs_NOT_RunningInTheUIThread();
ASSERT.Test(eds != null);
// Get the most similar PDFDocuments
int[] doc_ids = LDAAnalysisTools.GetDocumentsSimilarToDistribution(eds.LDAAnalysis, tags_distribution);
WPFDoEvents.InvokeInUIThread(() =>
{
WPFDoEvents.AssertThisCodeIsRunningInTheUIThread();
for (int i = 0; i < 10 && i < doc_ids.Length; ++i)
{
int doc_id = doc_ids[i];
string fingerprint = eds.docs[doc_id];
PDFDocument pdf_document = web_library_detail.Xlibrary.GetDocumentByFingerprint(fingerprint);
if (null == pdf_document)
{
Logging.Warn("Couldn't find similar document with fingerprint {0}", fingerprint);
}
else
{
PDFDocumentNodeContent content = new PDFDocumentNodeContent(pdf_document.Fingerprint, pdf_document.LibraryRef.Id);
NodeControlAddingByKeyboard.AddChildToNodeControl(node_control_, content, false);
}
}
});
}
void NodeControl_PreviewKeyDown(object sender, KeyEventArgs e)
{
//Logging.Info("NodeControl_PreviewKeyDown");
if (false)
{
}
else if (Key.Insert == e.Key)
{
NodeControl node_control = (NodeControl)sender;
if (KeyboardTools.IsCTRLDown())
{
object content = new StringNodeContent("Child node");
NodeControlAddingByKeyboard.AddChildToNodeControl(node_control, content);
}
else
{
NodeControlAddingByKeyboard.AddSiblingToNodeControl(node_control);
}
e.Handled = true;
}
}
private static void ExpandCitationsOutbound(PDFDocument doc, NodeControl node_control)
{
FeatureTrackingManager.Instance.UseFeature(Features.Brainstorm_ExploreLibrary_Document_CitationsOutbound);
ASSERT.Test(doc != null);
if (doc != null)
{
List <Citation> citations = doc.PDFDocumentCitationManager.GetOutboundCitations();
WPFDoEvents.InvokeInUIThread(() =>
{
WPFDoEvents.AssertThisCodeIsRunningInTheUIThread();
foreach (var citation in citations)
{
// NB: We assume the citations are from the same library!!
PDFDocumentNodeContent content = new PDFDocumentNodeContent(citation.fingerprint_inbound, doc.LibraryRef.Id);
if (!content.PDFDocument.Deleted)
{
NodeControlAddingByKeyboard.AddChildToNodeControl(node_control, content, false);
}
}
});
}
}
private static void ExpandAuthors(PDFDocument doc, NodeControl node_control)
{
WPFDoEvents.AssertThisCodeIs_NOT_RunningInTheUIThread();
ASSERT.Test(doc != null);
FeatureTrackingManager.Instance.UseFeature(Features.Brainstorm_ExploreLibrary_Document_Authors);
if (doc != null)
{
string authors = doc.AuthorsCombined;
if (String.IsNullOrEmpty(authors) || Constants.UNKNOWN_AUTHORS == authors)
{
return;
}
WPFDoEvents.InvokeInUIThread(() =>
{
WPFDoEvents.AssertThisCodeIsRunningInTheUIThread();
List <NameTools.Name> names = new List <NameTools.Name>();
string[] authors_split = NameTools.SplitAuthors_LEGACY(authors);
foreach (string author_split in authors_split)
{
string first_names, last_name;
NameTools.SplitName_LEGACY(author_split, out first_names, out last_name);
string initial = String.IsNullOrEmpty(first_names) ? null : first_names.Substring(0, 1);
PDFAuthorNodeContent pdf_author = new PDFAuthorNodeContent(doc.LibraryRef.Id, last_name, initial);
NodeControlAddingByKeyboard.AddChildToNodeControl(node_control, pdf_author, false);
}
});
}
}
private void ExpandAuthors()
{
FeatureTrackingManager.Instance.UseFeature(Features.Brainstorm_ExploreLibrary_Document_Authors);
PDFDocument pdf_document = pdf_document_node_content.PDFDocument;
string authors = pdf_document.AuthorsCombined;
if (String.IsNullOrEmpty(authors) || 0 == PDFDocument.UNKNOWN_AUTHORS.CompareTo(authors))
{
return;
}
List <NameTools.Name> names = new List <NameTools.Name>();
string[] authors_split = NameTools.SplitAuthors_LEGACY(authors);
foreach (string author_split in authors_split)
{
string first_names, last_name;
NameTools.SplitName_LEGACY(author_split, out first_names, out last_name);
string initial = String.IsNullOrEmpty(first_names) ? null : first_names.Substring(0, 1);
PDFAuthorNodeContent pdf_author = new PDFAuthorNodeContent(pdf_document.Library.WebLibraryDetail.Id, last_name, initial);
NodeControlAddingByKeyboard.AddChildToNodeControl(node_control, pdf_author, false);
}
}
private static void ExpandAnnotations(PDFDocument doc, NodeControl node_control)
{
WPFDoEvents.AssertThisCodeIs_NOT_RunningInTheUIThread();
ASSERT.Test(doc != null);
FeatureTrackingManager.Instance.UseFeature(Features.Brainstorm_ExploreLibrary_Document_Annotations);
if (doc != null)
{
var annotations = doc.GetAnnotations();
WPFDoEvents.InvokeInUIThread(() =>
{
WPFDoEvents.AssertThisCodeIsRunningInTheUIThread();
foreach (var annotation in annotations)
{
if (!annotation.Deleted)
{
PDFAnnotationNodeContent content = new PDFAnnotationNodeContent(doc.LibraryRef.Id, doc.Fingerprint, annotation.Guid.Value);
NodeControlAddingByKeyboard.AddChildToNodeControl(node_control, content, false);
}
}
});
}
}
private void ExpandDocuments()
{
FeatureTrackingManager.Instance.UseFeature(Features.Brainstorm_ExploreLibrary_Author_Documents);
List <PDFDocument> pdf_documents = SimilarAuthors.GetDocumentsBySameAuthorsSurnameAndInitial(pdf_author_node_content.LibraryRef, pdf_author_node_content.Surname, pdf_author_node_content.Initial);
foreach (PDFDocument pdf_document in pdf_documents)
{
PDFDocumentNodeContent content = new PDFDocumentNodeContent(pdf_document.Fingerprint, pdf_document.LibraryRef.Id);
NodeControlAddingByKeyboard.AddChildToNodeControl(node_control, content, false);
}
}
private void ExpandTags()
{
FeatureTrackingManager.Instance.UseFeature(Features.Brainstorm_ExploreLibrary_Document_Tags);
PDFDocument pdf_document = pdf_document_node_content.PDFDocument;
foreach (string tag in TagTools.ConvertTagBundleToTags(pdf_document.Tags))
{
PDFTagNodeContent pdf_tag = new PDFTagNodeContent(pdf_document.Library.WebLibraryDetail.Id, tag);
NodeControlAddingByKeyboard.AddChildToNodeControl(node_control, pdf_tag, false);
}
}
private void ExpandRelevants()
{
FeatureTrackingManager.Instance.UseFeature(Features.Brainstorm_ExploreLibrary_Document_Similars);
List <ExpeditionPaperSuggestions.Result> results = ExpeditionPaperSuggestions.GetRelevantOthers(pdf_document_node_content.PDFDocument, 10);
foreach (ExpeditionPaperSuggestions.Result result in results)
{
PDFDocumentNodeContent content = new PDFDocumentNodeContent(result.pdf_document.Fingerprint, result.pdf_document.Library.WebLibraryDetail.Id);
NodeControlAddingByKeyboard.AddChildToNodeControl(node_control, content, false);
}
}
public void ExpandDocuments()
{
FeatureTrackingManager.Instance.UseFeature(Features.Brainstorm_ExploreLibrary_Tag_Documents);
List <PDFDocument> pdf_documents = pdf_tag_node_content.Library.GetDocumentsByTag(pdf_tag_node_content.Tag);
foreach (PDFDocument pdf_document in pdf_documents)
{
PDFDocumentNodeContent content = new PDFDocumentNodeContent(pdf_document.Fingerprint, pdf_document.Library.WebLibraryDetail.Id);
NodeControlAddingByKeyboard.AddChildToNodeControl(node_control, content, false);
}
}
private void ExpandAnnotations()
{
FeatureTrackingManager.Instance.UseFeature(Features.Brainstorm_ExploreLibrary_Document_Annotations);
foreach (var annotation in pdf_document_node_content.PDFDocument.Annotations)
{
if (!annotation.Deleted)
{
PDFAnnotationNodeContent content = new PDFAnnotationNodeContent(pdf_document_node_content.PDFDocument.Library.WebLibraryDetail.Id, pdf_document_node_content.PDFDocument.Fingerprint, annotation.Guid.Value);
NodeControlAddingByKeyboard.AddChildToNodeControl(node_control, content, false);
}
}
}
private void ExpandAutoTags()
{
FeatureTrackingManager.Instance.UseFeature(Features.Brainstorm_ExploreLibrary_Document_AutoTags);
PDFDocument pdf_document = pdf_document_node_content.PDFDocument;
HashSet <string> tags = pdf_document.Library.AITagManager.AITags.GetTagsWithDocument(pdf_document.Fingerprint);
foreach (string tag in tags)
{
PDFAutoTagNodeContent pdf_auto_tag = new PDFAutoTagNodeContent(pdf_document.Library.WebLibraryDetail.Id, tag);
NodeControlAddingByKeyboard.AddChildToNodeControl(node_control, pdf_auto_tag, false);
}
}
private void ExpandCitationsInbound()
{
FeatureTrackingManager.Instance.UseFeature(Features.Brainstorm_ExploreLibrary_Document_CitationsInbound);
List <Citation> citations = pdf_document_node_content.PDFDocument.PDFDocumentCitationManager.GetInboundCitations();
foreach (var citation in citations)
{
PDFDocumentNodeContent content = new PDFDocumentNodeContent(citation.fingerprint_outbound, pdf_document_node_content.PDFDocument.Library.WebLibraryDetail.Id);
if (!content.PDFDocument.Deleted)
{
NodeControlAddingByKeyboard.AddChildToNodeControl(node_control, content, false);
}
}
}
public void ExpandDocuments()
{
FeatureTrackingManager.Instance.UseFeature(Features.Brainstorm_ExploreLibrary_AutoTag_Documents);
WebLibraryDetail web_library_detail = WebLibraryManager.Instance.GetLibrary(pdf_auto_tag_node_content.LibraryId);
HashSet <string> document_fingerprints = web_library_detail.Xlibrary.AITagManager.AITags.GetDocumentsWithTag(pdf_auto_tag_node_content.Tag);
List <PDFDocument> pdf_documents = web_library_detail.Xlibrary.GetDocumentByFingerprints(document_fingerprints);
foreach (PDFDocument pdf_document in pdf_documents)
{
PDFDocumentNodeContent content = new PDFDocumentNodeContent(pdf_document.Fingerprint, pdf_document.LibraryRef.Id);
NodeControlAddingByKeyboard.AddChildToNodeControl(node_control, content, false);
}
}
private static void ExpandThemes(PDFDocument doc, NodeControl node_control)
{
WPFDoEvents.AssertThisCodeIs_NOT_RunningInTheUIThread();
ASSERT.Test(doc != null);
FeatureTrackingManager.Instance.UseFeature(Features.Brainstorm_ExploreLibrary_Document_Themes);
if (doc != null)
{
ASSERT.Test(doc.LibraryRef.Xlibrary != null);
bool added_at_least_one_theme = false;
ExpeditionDataSource eds = doc.LibraryRef.Xlibrary?.ExpeditionManager?.ExpeditionDataSource;
if (null != eds)
{
if (eds.docs_index.ContainsKey(doc.Fingerprint))
{
int doc_id = eds.docs_index[doc.Fingerprint];
TopicProbability[] topics = eds.LDAAnalysis.DensityOfTopicsInDocsSorted[doc_id];
WPFDoEvents.InvokeInUIThread(() =>
{
for (int t = 0; t < Math.Min(topics.Length, 5); ++t)
{
string topic_name = eds.GetDescriptionForTopic(topics[t].topic, include_topic_number: false, "\n");
ThemeNodeContent tnc = new ThemeNodeContent(topic_name, doc.LibraryRef.Id);
NodeControlAddingByKeyboard.AddChildToNodeControl(node_control, tnc, false);
added_at_least_one_theme = true;
}
});
}
else
{
Logging.Warn("Expedition has not been run for library '{0}'.", doc.LibraryRef.Title);
}
}
if (!added_at_least_one_theme)
{
MessageBoxes.Warn("There were no themes available for this document. Please run Expedition against your library.");
}
}
}
internal void ExploreLibraryInBrainstorm(Library library)
{
BrainstormControl brainstorm_control = Instance.OpenNewBrainstorm();
int WIDTH = 320;
int HEIGHT = 240;
LibraryNodeContent content_library = new LibraryNodeContent(library.WebLibraryDetail.Title, library.WebLibraryDetail.Id);
NodeControl node_library = brainstorm_control.SceneRenderingControl.AddNewNodeControl(content_library, 0, 0, WIDTH, HEIGHT);
ExpeditionDataSource eds = library.ExpeditionManager.ExpeditionDataSource;
if (null != eds)
{
for (int topic = 0; topic < eds.lda_sampler.NumTopics; ++topic)
{
string topic_name = eds.GetDescriptionForTopic(topic, false, "\n");
ThemeNodeContent tnc = new ThemeNodeContent(topic_name, library.WebLibraryDetail.Id);
NodeControlAddingByKeyboard.AddChildToNodeControl(node_library, tnc);
}
}
else
{
{
StringNodeContent content_warning = new StringNodeContent("Please run Expedition on your library.");
NodeControl node_warning = brainstorm_control.SceneRenderingControl.AddNewNodeControl(content_warning, 0, -2 * HEIGHT);
brainstorm_control.SceneRenderingControl.AddNewConnectorControl(node_library, node_warning);
}
{
StringNodeContent content_warning = new StringNodeContent("Then you will get to\nexplore its themes.");
NodeControl node_warning = brainstorm_control.SceneRenderingControl.AddNewNodeControl(content_warning, -WIDTH, +2 * HEIGHT);
brainstorm_control.SceneRenderingControl.AddNewConnectorControl(node_library, node_warning);
}
{
StringNodeContent content_warning = new StringNodeContent("And you will get to\nexplore its documents.");
NodeControl node_warning = brainstorm_control.SceneRenderingControl.AddNewNodeControl(content_warning, +WIDTH, +2 * HEIGHT);
brainstorm_control.SceneRenderingControl.AddNewConnectorControl(node_library, node_warning);
}
}
brainstorm_control.AutoArrange = true;
}
internal static void AddDocumentsSimilarToDistribution(NodeControl node_control_, Library library, ExpeditionDataSource eds, float[] tags_distribution)
{
// Get the most similar PDFDocuments
int[] doc_ids = LDAAnalysisTools.GetDocumentsSimilarToDistribution(eds.LDAAnalysis, tags_distribution);
for (int i = 0; i < 10 && i < doc_ids.Length; ++i)
{
int doc_id = doc_ids[i];
string fingerprint = eds.docs[doc_id];
PDFDocument pdf_document = library.GetDocumentByFingerprint(fingerprint);
if (null == pdf_document)
{
Logging.Warn("Couldn't find similar document with fingerprint {0}", fingerprint);
}
else
{
PDFDocumentNodeContent content = new PDFDocumentNodeContent(pdf_document.Fingerprint, pdf_document.Library.WebLibraryDetail.Id);
NodeControlAddingByKeyboard.AddChildToNodeControl(node_control_, content, false);
}
}
}
private static void ExpandAutoTags(PDFDocument doc, NodeControl node_control)
{
WPFDoEvents.AssertThisCodeIs_NOT_RunningInTheUIThread();
ASSERT.Test(doc != null);
FeatureTrackingManager.Instance.UseFeature(Features.Brainstorm_ExploreLibrary_Document_AutoTags);
if (doc != null)
{
HashSet <string> tags = doc.LibraryRef.Xlibrary.AITagManager.AITags.GetTagsWithDocument(doc.Fingerprint);
WPFDoEvents.InvokeInUIThread(() =>
{
WPFDoEvents.AssertThisCodeIsRunningInTheUIThread();
foreach (string tag in tags)
{
PDFAutoTagNodeContent pdf_auto_tag = new PDFAutoTagNodeContent(doc.LibraryRef.Id, tag);
NodeControlAddingByKeyboard.AddChildToNodeControl(node_control, pdf_auto_tag, false);
}
});
}
}
private static void ExpandTags(PDFDocument doc, NodeControl node_control)
{
WPFDoEvents.AssertThisCodeIs_NOT_RunningInTheUIThread();
ASSERT.Test(doc != null);
FeatureTrackingManager.Instance.UseFeature(Features.Brainstorm_ExploreLibrary_Document_Tags);
if (doc != null)
{
var tags = TagTools.ConvertTagBundleToTags(doc.Tags);
WPFDoEvents.InvokeInUIThread(() =>
{
WPFDoEvents.AssertThisCodeIsRunningInTheUIThread();
foreach (string tag in tags)
{
PDFTagNodeContent pdf_tag = new PDFTagNodeContent(doc.LibraryRef.Id, tag);
NodeControlAddingByKeyboard.AddChildToNodeControl(node_control, pdf_tag, false);
}
});
}
}
private static void ExpandRelevants(PDFDocument doc, NodeControl node_control)
{
WPFDoEvents.AssertThisCodeIs_NOT_RunningInTheUIThread();
ASSERT.Test(doc != null);
FeatureTrackingManager.Instance.UseFeature(Features.Brainstorm_ExploreLibrary_Document_Similars);
if (doc != null)
{
List <ExpeditionPaperSuggestions.Result> results = ExpeditionPaperSuggestions.GetRelevantOthers(doc, 10);
WPFDoEvents.InvokeInUIThread(() =>
{
WPFDoEvents.AssertThisCodeIsRunningInTheUIThread();
foreach (ExpeditionPaperSuggestions.Result result in results)
{
PDFDocumentNodeContent content = new PDFDocumentNodeContent(result.pdf_document.Fingerprint, result.pdf_document.LibraryRef.Id);
NodeControlAddingByKeyboard.AddChildToNodeControl(node_control, content, false);
}
});
}
}
internal static void AddDocumentsInfluentialInDistribution(NodeControl node_control_, Library library, ExpeditionDataSource eds, float[] tags_distribution)
{
Logging.Info("+Performing ThemedPageRank on {0} documents", eds.LDAAnalysis.NUM_DOCS);
// We have the distribution of the topic in tags_distribution
// Create an array for the document biases
// Fill the array using the dot product of the document distribution dotted with the topic distribution - then normalise
double[] biases = new double[eds.LDAAnalysis.NUM_DOCS];
for (int doc = 0; doc < eds.LDAAnalysis.NUM_DOCS; ++doc)
{
double bias_num_squared = 0;
double bias_den_doc = 0;
double bias_den_tags = 0;
for (int topic = 0; topic < eds.LDAAnalysis.NUM_TOPICS; ++topic)
{
bias_num_squared += eds.LDAAnalysis.DensityOfTopicsInDocuments[doc, topic] * tags_distribution[topic];
bias_den_doc += eds.LDAAnalysis.DensityOfTopicsInDocuments[doc, topic] * eds.LDAAnalysis.DensityOfTopicsInDocuments[doc, topic];
bias_den_tags += tags_distribution[topic] * tags_distribution[topic];
}
biases[doc] = bias_num_squared / (Math.Sqrt(bias_den_doc) * Math.Sqrt(bias_den_tags));
}
// Then build up a matrix FROM each document -
List <int>[] references_outbound = new List <int> [eds.LDAAnalysis.NUM_DOCS];
for (int doc = 0; doc < eds.LDAAnalysis.NUM_DOCS; ++doc)
{
references_outbound[doc] = new List <int>();
string fingerprint = eds.docs[doc];
PDFDocument pdf_document = library.GetDocumentByFingerprint(fingerprint);
if (null == pdf_document)
{
Logging.Warn("ThemeExplorer::AddInInfluential: Cannot find document anymore for fingerprint {0}", fingerprint);
}
else
{
List <Citation> citations_outbound = pdf_document.PDFDocumentCitationManager.GetOutboundCitations();
foreach (Citation citation in citations_outbound)
{
string fingerprint_inbound = citation.fingerprint_inbound;
if (eds.docs_index.ContainsKey(fingerprint_inbound))
{
int doc_inbound = eds.docs_index[fingerprint_inbound];
references_outbound[doc].Add(doc_inbound);
}
}
}
}
// Space for the pageranks
double[] pageranks_current = new double[eds.LDAAnalysis.NUM_DOCS];
double[] pageranks_next = new double[eds.LDAAnalysis.NUM_DOCS];
// Initialise
for (int doc = 0; doc < eds.LDAAnalysis.NUM_DOCS; ++doc)
{
pageranks_current[doc] = biases[doc];
}
// Iterate
int NUM_ITERATIONS = 20;
for (int iteration = 0; iteration < NUM_ITERATIONS; ++iteration)
{
Logging.Info("Performing ThemedPageRank iteration {0}", iteration);
// Spread out the activation pageranks
for (int doc = 0; doc < eds.LDAAnalysis.NUM_DOCS; ++doc)
{
foreach (int doc_inbound in references_outbound[doc])
{
pageranks_next[doc_inbound] += biases[doc] / references_outbound[doc].Count;
}
}
// Mix the spread out pageranks with the initial bias pageranks
double ALPHA = 0.5;
for (int doc = 0; doc < eds.LDAAnalysis.NUM_DOCS; ++doc)
{
pageranks_next[doc] = (1 - ALPHA) * pageranks_next[doc] + ALPHA * biases[doc];
}
// Normalise the next pageranks
double total = 0;
for (int doc = 0; doc < eds.LDAAnalysis.NUM_DOCS; ++doc)
{
total += pageranks_next[doc];
}
if (0 < total)
{
for (int doc = 0; doc < eds.LDAAnalysis.NUM_DOCS; ++doc)
{
pageranks_next[doc] /= total;
}
}
// Switch in the next pageranks because we will overwrite them
double[] pageranks_temp = pageranks_current;
pageranks_current = pageranks_next;
pageranks_next = pageranks_temp;
}
// Sort the pageranks, descending
int[] docs = new int[eds.LDAAnalysis.NUM_DOCS];
for (int doc = 0; doc < eds.LDAAnalysis.NUM_DOCS; ++doc)
{
docs[doc] = doc;
}
Array.Sort(pageranks_current, docs);
Array.Reverse(pageranks_current);
Array.Reverse(docs);
// Make the nodes
for (int doc = 0; doc < 10 && doc < docs.Length; ++doc)
{
int doc_id = docs[doc];
string fingerprint = eds.docs[doc_id];
PDFDocument pdf_document = library.GetDocumentByFingerprint(fingerprint);
if (null == pdf_document)
{
Logging.Warn("Couldn't find similar document with fingerprint {0}", fingerprint);
}
else
{
PDFDocumentNodeContent content = new PDFDocumentNodeContent(pdf_document.Fingerprint, pdf_document.Library.WebLibraryDetail.Id);
NodeControlAddingByKeyboard.AddChildToNodeControl(node_control_, content, false);
}
}
}
