Dwi Hendratmo Widyantoro
Institut Teknologi Bandung

Published : 11 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 11 Documents
Search

EXPLOITING UNLABELED DATA IN CONCEPT DRIFT LEARNING Widyantoro, Dwi Hendratmo
Jurnal Informatika Vol 8, No 1 (2007): MAY 2007
Publisher : Institute of Research and Community Outreach - Petra Christian University

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (183.665 KB) | DOI: 10.9744/informatika.8.1.pp. 54-62

Abstract

Learning unlabeled data in a drifting environment still receives little attention. This paper presents a concept tracker algorithm for learning concept drift that exploits unlabeled data. In the absence of complete labeled data, instance classes are identified using a concept hierarchy that is incrementally constructed from data stream (mostly unlabeled data) in unsupervised mode. The persistence assumption in temporal reasoning is then applied to infer target concepts. Empirical evaluation that has been conducted on information-filtering domains demonstrates the effectiveness of this approach.
Shared-hidden-layer Deep Neural Network for Under-resourced Language the Content Devin Hoesen; Dessi Puji Lestari; Dwi Hendratmo Widyantoro
TELKOMNIKA (Telecommunication Computing Electronics and Control) Vol 16, No 3: June 2018
Publisher : Universitas Ahmad Dahlan

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12928/telkomnika.v16i3.7984

Abstract

Training speech recognizer with under-resourced language data still proves difficult. Indonesian language is considered under-resourced because the lack of a standard speech corpus, text corpus, and dictionary. In this research, the efficacy of augmenting limited Indonesian speech training data with highly-resourced-language training data, such as English, to train Indonesian speech recognizer was analyzed. The training was performed in form of shared-hidden-layer deep-neural-network (SHL-DNN) training. An SHL-DNN has language-independent hidden layers and can be pre-trained and trained using multilingual training data without any difference with a monolingual deep neural network. The SHL-DNN using Indonesian and English speech training data proved effective for decreasing word error rate (WER) in decoding Indonesian dictated-speech by achieving 3.82% absolute decrease compared to a monolingual Indonesian hidden Markov model using Gaussian mixture model emission (GMM-HMM). The case was confirmed when the SHL-DNN was also employed to decode Indonesian spontaneous-speech by achieving 4.19% absolute WER decrease.
The Strategies for Quorum Satisfaction in Host-to-Host Meeting Scheduling Negotiation Rani Megasari; Kuspriyanto Kuspriyanto; Emir Mauludi Husni; Dwi Hendratmo Widyantoro
TELKOMNIKA (Telecommunication Computing Electronics and Control) Vol 14, No 4: December 2016
Publisher : Universitas Ahmad Dahlan

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12928/telkomnika.v14i4.4521

Abstract

This paper proposes two strategies for handling conflict schedule of two meetings which invite the same member of personnel at the same time through host-to-host negotiation scheme. The strategy is to let the member attend the other meeting under the condition that the group decision regarding the schedule is not changed and meeting quorum is fulfilled, namely release strategy. Other strategy is to substitute the absent personnel in order to keep the number of attendees above the quorum, namely substitute strategy. This paper adapts a mechanism design approach, namely Clarke Tax Mechanism, to satisfy incentive compatibility and individual rationality principal in meeting scheduling. By using a release strategy and substitute strategy, colliding meetings can still be held according to the schedule without the need for rescheduling. This paper shows the simulation result of using the strategies within some scenarios. It demonstrates that the number of meeting failures can be reduced with negotiation.        
Towards host-to-host meeting scheduling negotiation Rani Megasari; Kuspriyanto Kuspriyanto; Emir Mauludi Husni; Dwi Hendratmo Widyantoro
International Journal of Advances in Intelligent Informatics Vol 1, No 1 (2015): March 2015
Publisher : Universitas Ahmad Dahlan

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26555/ijain.v1i1.6

Abstract

This paper presents a different scheme of meeting scheduling negotiation among a large number of personnel in a heterogeneous community. This scheme, named Host-to-Host Negotiation, attempts to produce a stable schedule under uncertain personnel preferences. By collecting information from hosts’ inter organizational meeting, this study intends to guarantee personnel availability. As a consequence, personnel’s and meeting’s profile in this scheme are stored in a centralized manner. This study considers personnel preferences by adapting the Clarke Tax Mechanism, which is categorized as a non manipulated mechanism design. Finally, this paper introduces negotiation strategies based on the conflict handling mode. A host-to-host scheme can give notification if any conflict exist and lead to negotiation process with acceptable disclosed information. Nevertheless, a complete negotiation process will be more elaborated in the future works.
TRANSFORMING RHETORICAL DOCUMENT PROFILE INTO TAILORED SUMMARY OF SCIENTIFIC PAPER Masayu Leylia Khodra; Mohammad Dimas; Dwi Hendratmo Widyantoro; E. Aminudin Aziz; Bambang Riyanto Trilaksono
Jurnal Ilmiah Kursor Vol 6 No 3 (2012)
Publisher : Universitas Trunojoyo Madura

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Since abstract of scientific paper is author biased, readers’ required information may not be included in the abstract. Tailored summary may help them to get a summary based on their information needs. This research is the first one that implements tailored summary system for scientific paper. Tailored summary applies information extraction that transforms a scientific paper into Rhetorical Document Profile, a structured representation of paper content based on rhetorical scheme of fifteen slots. This research adapted building plan that used rhetorical scheme of seven slots. We also implement tailored summary system. After generating initial summary, surface repair is conducted to improve summary readability. Each sentence in initial summary is combined with template phrase based on syntax-tree combination method. There are five groups of template phrases provided in surface repair. We construct evaluation standards by asking five human raters. The best method for sentence selection subsystem that uses Maximal Marginal Importance-Multi Sentence is employing TF.IDF weighting system with precision/recall of 0.61. The surface repair subsystem has acceptance of 0.91.
Automatic Tailored Multi-Paper Summarization based on Rhetorical Document Profile and Summary Specification Masayu Leylia Khodra; Dwi Hendratmo Widyantoro; E. Aminudin Aziz; Bambang Riyanto Trilaksono
Journal of ICT Research and Applications Vol. 6 No. 3 (2012)
Publisher : LPPM ITB

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.5614/itbj.ict.2012.6.3.4

Abstract

In  order  to  assist  researchers  in  addressing  time  constraint  and  low relevance  in  using  scientific  articles,  an  automatic  tailored  multi-paper summarization  (TMPS)  is  proposed.  In  this  paper,  we  extend  Teufel's  tailored summary  to  deal  with  multi-papers  and  more  flexible  representation  of  user information needs. Our TMPS extracts Rhetorical Document Profile (RDP) from each paper and  presents a summary based on user information needs.  Building Plan  Language  (BPLAN)  is  introduced  as  a  formalization  of  Teufel's  building plan  and  used  to  represent summary  specification,  which  is  more  flexible representation user information needs. Surface repair is embedded within the BPLAN  for  improving  the  readability  of  extractive summary.  Our  experiment shows that the average performance of RDP extraction module is 94.46%, which promises  high  quality  of  extracts  for  summary  composition.  Generality evaluation  shows  that  our  BPLAN  is  flexible  enough  in  composing  various forms  of summary.  Subjective  evaluation  provides evidence that  surface repair operators can improve the resulting summary readability.
Word Embedding for Rhetorical Sentence Categorization on Scientific Articles Ghoziyah Haitan Rachman; Masayu Leylia Khodra; Dwi Hendratmo Widyantoro
Journal of ICT Research and Applications Vol. 12 No. 2 (2018)
Publisher : LPPM ITB

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.5614/itbj.ict.res.appl.2018.12.2.5

Abstract

A common task in summarizing scientific articles is employing the rhetorical structure of sentences. Determining rhetorical sentences itself passes through the process of text categorization. In order to get good performance, some works in text categorization have been done by employing word embedding. This paper presents rhetorical sentence categorization of scientific articles by using word embedding to capture semantically similar words. A comparison of employing Word2Vec and GloVe is shown. First, two experiments are evaluated using five classifiers, namely Naïve Bayes, Linear SVM, IBK, J48, and Maximum Entropy. Then, the best classifier from the first two experiments was employed. This research showed that Word2Vec CBOW performed better than Skip-Gram and GloVe. The best experimental result was from Word2Vec CBOW for 20,155 resource papers from ACL-ARC, features from Teufel and the previous label feature. In this experiment, Linear SVM produced the highest F-measure performance at 43.44%.
Efficient Utilization of Dependency Pattern and Sequential Covering for Aspect Extraction Rule Learning Fariska Zakhralativa Ruskanda; Dwi Hendratmo Widyantoro; Ayu Purwarianti
Journal of ICT Research and Applications Vol. 14 No. 1 (2020)
Publisher : LPPM ITB

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.5614/itbj.ict.res.appl.2020.14.1.4

Abstract

The use of dependency rules for aspect extraction tasks in aspect-based sentiment analysis is a promising approach. One problem with this approach is incomplete rules. This paper presents an aspect extraction rule learning method that combines dependency rules with the Sequential Covering algorithm. Sequential Covering is known for its characteristics in constructing rules that increase positive examples covered and decrease negative ones. This property is vital to make sure that the rule set used has high performance, but not inevitably high coverage, which is a characteristic of the aspect extraction task. To test the new method, four datasets were used from four product domains and three baselines: Double Propagation, Aspectator, and a previous work by the authors. The results show that the proposed approach performed better than the three baseline methods for the F-measure metric, with the highest F-measure value at 0.633.
Analisis Pembangunan Korpus Berpasangan Untuk Pembangkitan Parafrasa Pada Makalah Ilmiah Ridwan Ilyas; Dwi Hendratmo Widyantoro; Masayu Leylia Khodra
JUMANJI (Jurnal Masyarakat Informatika Unjani) Vol 2 No 1 (2018): Jurnal Masyarakat Informatika Unjani
Publisher : Jurusan Informatika Universitas Jenderal Achmad Yani

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (154.731 KB) | DOI: 10.26874/jumanji.v2i1.44

Abstract

Pembangunan mesin yang dapat membangkitkan kalimat baru dengan tingkat semantik yang tinggi namun secara penulisan berbeda (parafrasa) membutuhkan sumberdaya bahasa berupa korpus parallel. Proses pembangunan korpus memerlukan analisis awal sesuai dengan domain dari mesin yang akan dibuat. Pada penelitian ini dilakukan analis dalam pembangunan korpus berpasangan pada makalah ilmiah. Kalimat-kalimat pada makalah ilmiah memiliki karakteristik yang berbeda dengan domain lain seperti berita atau media sosial. Dari hasil proses ekstraksi awal didapatkan 590.402 kalimat isi dan 23.584 kalimat abstrak. Hasil dari penelitian ini dapat menjadi kandidat korpus yang dilakukan dengan proses terkomputerisasi.
Winner-Takes-All based Multi-Strategy Learning for Information Extraction Dwi Hendratmo Widyantoro; Kurnia Muludi; Kuspriyanto Kuspriyanto
Indonesian Journal of Electrical Engineering and Computer Science Vol 12, No 11: November 2014
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/ijeecs.v12.i11.pp7935-7945

Abstract

This paper proposes a winner-takes-all based multi-strategy learning for information extraction. Unlike the majority of multi-strategy approaches that commonly combine the prediction of all base learnings involved, our approach takes a different strategy by employing only the best, single predictor for a specific information task. The best predictor (among other predictors) is identified during training phase using k-fold cross validation, which is then retrained on the full training set. Empirical evaluation on two benchmarks data sets demonstrates the effectiveness of our strategy. Out of 26 information extraction cases, our strategy outperforms other information extraction algorithms and strategies in 16 cases. The winner-takes-all strategy in general eliminates the difficult situation in multi-strategy learning when the majority of base learners cannot make correct prediction, resulting in incorrect prediction on its output. In such a case, the best predictor with correct prediction  in our strategy will take over for the overal prediction.