A novel approach to capture the similarity in summarized text using embedded model

Mishra, Asha Rani; Panchal, Vishal

doi:10.2478/ijssis-2022-0002

Cited by 4 publications

(4 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In the field of text summarization, text matching techniques are also utilized to identify duplicate text content, facilitating the removal of redundant information from summaries, which is beneficial for text mining purposes. Mishra et al [33] proposed an embedded model to examine the similarity of summary texts, effectively addressing this issue. Long-text QA is also an important challenge.…”

Section: Text Matchingmentioning

confidence: 99%

FinKENet: A Novel Financial Knowledge Enhanced Network for Financial Question Matching

Guo,

Liang,

Chen

et al. 2023

Entropy

View full text Add to dashboard Cite

Question matching is the fundamental task in retrieval-based dialogue systems which assesses the similarity between Query and Question. Unfortunately, existing methods focus on improving the accuracy of text similarity in the general domain, without adaptation to the financial domain. Financial question matching has two critical issues: (1) How to accurately model the contextual representation of a financial sentence? (2) How to accurately represent financial key phrases in an utterance? To address these issues, this paper proposes a novel Financial Knowledge Enhanced Network (FinKENet) that significantly injects financial knowledge into contextual text. Specifically, we propose a multi-level encoder to extract both sentence-level features and financial phrase-level features, which can more accurately represent sentences and financial phrases. Furthermore, we propose a financial co-attention adapter to combine sentence features and financial keyword features. Finally, we design a multi-level similarity decoder to calculate the similarity between queries and questions. In addition, a cross-entropy-based loss function is presented for model optimization. Experimental results demonstrate the effectiveness of the proposed method on the Ant Financial question matching dataset. In particular, the Recall score improves from 73.21% to 74.90% (1.69% absolute).

show abstract

Section: Text Matchingmentioning

confidence: 99%

FinKENet: A Novel Financial Knowledge Enhanced Network for Financial Question Matching

Guo,

Liang,

Chen

et al. 2023

Entropy

View full text Add to dashboard Cite

show abstract

“…Setelah didapatkan nilai shingle dokumen, kemudian dengan Cosine Similarity akan dihitung nilai kemiripan dari dua buah dokumen berdasarkan vektor nilai shingle dokumen tersebut. Nilai Cosine Similarity ini menandakan persentase kemiripan masing-masing dokumen, dimana terdapat nilai ambang batas (threshold) yang digunakan untuk menentukan apakah dua buah dokumen terindikasi melakukan plagiat, rentang nilai tersebut adalah < 60%, 60 -70%, 70 -80%, dan > 80% [17].…”

Section: Gambar 3: Flowchart Pengembangan Fitur Deteksi Kemiripan Dok...unclassified

“…Nilai K-Shingling yang dihasilkan dari masing-masing dokumen menandakan apakah shingle tersebut ditemukan pada dokumen bersangkutan, apabila shingle tersebut ditemukan, maka nilai nya 1, sedangkan jika shingle tidak ditemukan pada dokumen, maka nilai nya 0. Setelah didapatkan nilai K-Shingling dari kedua dokumen, kemiripan dokumen dapat diukur berdasarkan proporsi nilai shingle yang ditemukan dalam pasangan dokumen teks dari vektor kata yang didefinisikan untuk mewakili dokumen [17], dan kemudian vektor tersebut akan dibandingkan menggunakan Cosine Similarity Dengan menggunakan rumus pada "(2)", maka didapat nilai sebagai berikut :…”

Section: A Hasil Implementasi Metodeunclassified

Perancangan Fitur Deteksi Kemiripan Dokumen Jawaban Tugas Mahasiswa pada Sistem Manajemen Pembelajaran dengan Metode K-Shingling dan Cosine Similarity

Artawan¹,

Hartati²,

Divayana³

2023

JTE

View full text Add to dashboard Cite

Perkembangan teknologi dapat digunakan untuk mendukung proses belajar mengajar agar berjalan dengan lebih efektif dan efisien, salah satu aplikasi yang dapat digunakan untuk menunjang hal tersebut adalah Sistem Manajemen Pembelajaran atau yang sering dikenal dengan LMS (Learning Management System). LMS menyediakan wadah yang terintegrasi dimulai dari penyampaian materi hingga evaluasi pembelajaran. Metode evaluasi pembelajaran yang dapat dilakukan pada LMS adalah dengan pemberian tugas kepada mahasiswa, dan mahasiswa dapat melakukan pengumpulan dengan mengunggah dokumen jawaban mereka. Hal yang sering menjadi permasalahan adalah adanya indikasi plagiarisme yang dilakukan mahasiswa dengan menduplikasi jawaban mahasiswa lainnya. Untuk mengatasi permasalahan tersebut, akan dikembangkan fitur pada menu Tugas Mata Kuliah LMS yang dapat mendeteksi kemiripan dokumen jawaban tugas mahasiswa dengan menggunakan metode K-Shingling dan Cosine Similarity. Algoritma K-Shingling digunakan untuk membentuk shingle yang merupakan substring atau potongan kata yang dibentuk dari keseluruhan teks dalam dokumen jawaban tugas yang telah melalui tahap preprocessing, kemudian untuk tiap dokumen dibangkitkan suatu vektor nilai shingle yang akan digunakan untuk membandingkan nilai kemiripan antar dokumen menggunakan Cosine Similarity. Hasil yang didapat dalam penelitian ini adalah sistem yang dikembangkan mampu mendeteksi persentase kemiripan dokumen jawaban tugas dari tiap mahasiswa, dan dari nilai persentase tersebut dapat mengindikasikan apakah jawaban tugas mahasiswa merupakan plagiat terhadap jawaban mahasiswa lainnya atau tidak. Kata Kunci— LMS; Kemiripan Dokumen; K-Shingling; Cosine Similarity

show abstract

“…The phase of document selection is employed to identify the most pertinent sections in relation to the stipulated learning objective. This is accomplished through a similarity search using cosine similarity, yielding the top n-documents, with n representing a parameter defined by the user (Mishra et al, 2020). This process employs the learning goal sentence, which is vectorized using embeddings.…”

Section: Technical Parts Of the Proposed Systemmentioning

confidence: 99%

Generative Agents to support students learning progress

Schacht,

Kamath Barkur,

Lanquillon

2023

Libro De Actas.

View full text Add to dashboard Cite

Ongoing assessments in a course are crucial for tracking student performance and progress. However, generating and evaluating tests for each lesson and student can be time-consuming. Existing models for generating and evaluating question-answer pairs have had limited success. In recent years, large language models (LLMs) have become available as a service, offering more intelligent answering and evaluation capabilities. This research aims to leverage LLMs for generating questions, model answers, and evaluations while providing valuable feedback to students and decentralizing the dependency on faculty.We finetune existing LLMs and employ prompt engineering to direct the model toward specific tasks using different generative agents. One agent generates questions, another generates answers, and the third takes the human answers to evaluate and ensure the quality. Human evaluation is conducted through focus group analysis, and student progress and faculty feedback are tracked. Results demonstrate the potential of LLMs to provide automatic feedback and learning progress tracking for both students and faculty.In conclusion, this paper demonstrates the versatility of LLMs for various learning tasks, including question generation, model answer generation, and evaluation, all while providing personalized feedback to students. By identifying and addressing knowledge gaps, LLMs can support continuous evaluation and help students improve their understanding before semester exams. Furthermore, knowledge gaps from students identified by the agent can be highlighted and addressed through additional classes or support materials, potentially generated by the same model, leading to a more personalized learning experience.

show abstract

A novel approach to capture the similarity in summarized text using embedded model

Cited by 4 publications

References 37 publications

FinKENet: A Novel Financial Knowledge Enhanced Network for Financial Question Matching

FinKENet: A Novel Financial Knowledge Enhanced Network for Financial Question Matching

Perancangan Fitur Deteksi Kemiripan Dokumen Jawaban Tugas Mahasiswa pada Sistem Manajemen Pembelajaran dengan Metode K-Shingling dan Cosine Similarity

Generative Agents to support students learning progress

Contact Info

Product

Resources

About