Explainable classification
Inspect a deterministic decision
Select a business document. The server extracts its structure, scores every profile, and returns the evidence behind the winning category.
DOCX SAMPLE
Document to classify
LOCAL
Amounts, invoice identifiers, billing details, and payment terms.
PREDICTED CATEGORY
Pending
0Profiles
0Signals
0 msServer time
SCOREBOARD
Competing categories
WEIGHTEDEXPLANATION
Strongest evidence
TRACEReady to tryUse the controls above to interact with this demo.
What this demonstrates
Classify documents without a model or external service
A deterministic classifier always produces the same answer for the same document and profile configuration. Every score can be traced to a rule, its document region, and its weighted contribution.
Key members
Load DOCX
ServerTextControl.Load
Extract body structure
ServerTextControl.Paragraphs
Inspect repeated content
Section.HeadersAndFooters
Show the selected DOCX
TXDocumentViewer.loadDocument
How it works
Classification profiles are loaded from JSON. Each rule defines a term, match mode, weight, strength, and optional target region, allowing domain experts to tune behavior without retraining or changing application code.
TX Text Control reads paragraphs plus section headers and footers. Paragraph styles distinguish titles and headings from ordinary body content, so important structural signals receive more weight.
Repeated matches are damped with a square root, early segments receive a modest position boost, and confidence combines winner separation with rule coverage. The result includes the complete category scoreboard and the strongest contributing matches.
var contribution = Math.Sqrt(weightedOccurrences)
* rule.Weight
* RegionWeight(region)
* KeywordStrengthWeight(rule);
Technique adapted from Document Classification Without AI.
NuGet packages
TXTextControl.TextControl.Core.SDK
34.0.4
Provides the core document model and server-side document processing APIs.
dotnet add package TXTextControl.TextControl.Core.SDK --version 34.0.4
Displays supported document and PDF formats in the browser.
dotnet add package TXTextControl.Web.DocumentViewer --version 34.2.0
