International Conference on Computer and Knowledge Engineering

Home / 15th International Conference on Computer and Knowledge Engineering

A Hybrid Architecture to Optimize Persian FAQ Retrieval using Semantic Similarity Search

Authors :

Seyed Amir Mohammad Hosseini¹ Fatemeh Dehbashi² Setare Kahnemuee³ Mohsen Kahani⁴ Morteza Fardin⁵

1- Computer Engineering Department Ferdowsi University of Mashhad 2- Computer Engineering Department Ferdowsi University of Mashhad 3- Computer Engineering Department Ferdowsi University of Mashhad 4- Computer Engineering Department Ferdowsi University of Mashhad 5- Baran Software Group

Keywords :

Persian Language Processing،FAQ Retrieval،Hybrid Clustering،Sentence Embeddings،Semantic Search،Evaluation Framework

Abstract :

Traditional FAQ-based Question Answering (QA) systems, which rely on lexical matching, often fail to comprehend the semantic nuances of morphologically rich languages like Persian. Furthermore, their performance is typically measured by rigid metrics that underestimate a model's true capabilities. This paper addresses these challenges by proposing a novel, two-stage Hybrid Clustering-Based Search architecture designed to improve both the accuracy and efficiency of semantic retrieval. We introduce an iterative "Hierarchical DBSCAN" method to cluster a real-world Persian knowledge base, allowing for a focused, coarse-to-fine search pipeline. To robustly evaluate our system, we created a new multi-faceted dataset containing formal, informal, and challenging queries. Our experiments show that the proposed hybrid architecture, achieves a Top-1 accuracy of 79% and a Top-3 accuracy of 86% and also answering delay of only 0.115 seconds. This represents an improvement in both accuracy and speed compared to the optimal E5-Base model in the standard Direct Semantic Search baseline. Our work provides an efficient and validated blueprint for developing practical semantic QA systems for the Persian language.

Papers List

List of archived papers

Underwater Image Super-Resolution using Generative Adversarial Network-based Model

Alireza Aghelan - Modjtaba Rouhani

Automatic Generation of XACML Code using Model-Driven Approach

Athareh Fatemian - Bahman Zamani - Marzieh Masoumi - Mehran Kamranpour - Behrouz Tork Ladani - Shekoufeh Kolahdouz Rahimi

Optimizing MR Image Registration for Accurate Brain Volume Measurement in Children with Autism Spectrum Disorder

Shiva Sanati - Mahdi Saadatmand

FGM Copula based Analysis of Coverage Region for Wireless Three-User Multiple Access Channel with Correlated Channel Coefficients

Mona Sadat Mohsenzadeh - Ghosheh Abed Hodtani

A Novel Approach for Image-Text Matching Cross-Modal Space Learning

Amirreza Ebrahimi - Mohammad Javad Parseh - Pejman Rasti

Enhancing Cloud Security with Federated CNN-LSTM: A Novel Approach to Intrusion Detection

Reyhaneh Ilaghi - Raheleh Ilaghi - Fereshteh Rahmani - Seyyed hamid Ghafoori

Adaptive Pronunciation Scoring: Aligning Automated Assessments with Human Expert Evaluations

Omid Aghdaei - Mohammad Sadegh Safari - Mohammad Hassan Rasoolizadeh - Abedeh Mirzaee

Diagnosis of Depression Based on New Features Extractive from the Frequency Space of the EEG

Melika Changizi - Saeid Rashidi

Improving ADHD Detection with Cost-Sensitive LightGBM

Behnam Yousefimehr - Mehdi Ghatee - Ali Heydari

LightFedSelect: A Lightweight Framework for Byzantine-Robust Federated Learning

Seyed Saeed Razavi - Seyed Arsalan Vasegh Rahim Parvar - Soroosh Dadashi Pakdeh - Mohammad Matin Rezaeifard - Morteza Mollaie Chafi - Reza Ebrahimi Atani