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 Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Question-answering systems and voice assistants are becoming major part ofclient service departments of many organizations, helping them to reduce thelabor costs of staff. In many such systems, there is always natural languageunderstanding module that solves intent classification task. This task iscomplicated because of its case-dependency - every subject area has its ownsemantic kernel. The state of art approaches for intent classification aredifferent machine learning and deep learning methods that use text vectorrepresentations as input. The basic vector representation models such as Bag ofwords and TF-IDF generate sparse matrixes, which are becoming very big as theamount of input data grows. Modern methods such as word2vec and FastText useneural networks to evaluate word embeddings with fixed dimension size. As weare developing a question-answering system for students and enrollees of thePerm National Research Polytechnic University, we have faced the problem ofuser's intent detection. The subject area of our system is very specific, thatis why there is a lack of training data. This aspect makes intentclassification task more challenging for using state of the art deep learningmethods. In this paper, we propose an approach of the questions embeddingsrepresentation based on calculation of Shannon entropy.The goal of the approachis to produce low dimensional question vectors as neural approaches do and tooutperform related methods, described above in condition of small dataset. Weevaluate and compare our model with existing ones using logistic regression anddataset that contains questions asked by students and enrollees. The data islabeled into six classes. Experimental comparison of proposed approach andother models revealed that proposed model performed better in the given task.

Question Embeddings Based on Shannon Entropy: Solving intent classification task in goal-oriented dialogue system