کنفرانس بین المللی فناوری اطلاعات و دانش

صفحه اصلی / شانزدهمین کنفرانس بین المللی فناوری اطلاعات و دانش

HTCAR: Hierarchical Text Classification based on aggregation of Representations

نویسندگان :

Ali Bavand¹ Mohammad Mehdi Homayounpour² Ahmad Nickabadi³

1- صنعتی امی 2- دانشگاه‌ صنعتی امیرکبیر (پلی‌تکنیک تهران) 3- دانشگاه‌ صنعتی امیرکبیر (پلی‌تکنیک تهران)

کلمات کلیدی :

Hierarchical text classification،text processing،text representation،word embedding،financial news

چکیده :

Hierarchical Text Classification (HTC) involves assigning labels with hierarchical structures to text data. This presents challenges due to complex relationships between labels and the need for accurate representation of these interactions. Existing methods often fail to adequately address this challenge. This research introduces a novel HTC method based on aggregating representations to enhance classification accuracy. This approach was specifically designed to address the challenges of classifying financial news (in Persian dataset), where accurate analysis is crucial for informed decision-making. Evaluated on both English (WOS) and Persian (Hamshahri Online) datasets, the proposed method achieved a 1–2% improvement in classification accuracy, demonstrating its effectiveness across different languages and datasets. This improvement highlights the model's effectiveness in handling the complexities of hierarchical text classification.