English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Text documents can be described by a number of abstract concepts such assemantic category, writing style, or sentiment. Machine learning (ML) modelshave been trained to automatically map documents to these abstract concepts,allowing to annotate very large text collections, more than could be processedby a human in a lifetime. Besides predicting the text's category veryaccurately, it is also highly desirable to understand how and why thecategorization process takes place. In this paper, we demonstrate that suchunderstanding can be achieved by tracing the classification decision back toindividual words using layer-wise relevance propagation (LRP), a recentlydeveloped technique for explaining predictions of complex non-linearclassifiers. We train two word-based ML models, a convolutional neural network(CNN) and a bag-of-words SVM classifier, on a topic categorization task andadapt the LRP method to decompose the predictions of these models onto words.Resulting scores indicate how much individual words contribute to the overallclassification decision. This enables one to distill relevant information fromtext documents without an explicit semantic information extraction step. Wefurther use the word-wise relevance scores for generating novel vector-baseddocument representations which capture semantic information. Based on thesedocument vectors, we introduce a measure of model explanatory power and showthat, although the SVM and CNN models perform similarly in terms ofclassification accuracy, the latter exhibits a higher level of explainabilitywhich makes it more comprehensible for humans and potentially more useful forother applications.

"What is Relevant in a Text Document?": An Interpretable Machine Learning Approach