The research paper published by IJSER journal is about Issues and challenges of Research: Originality Based Document Classification and Information Extraction 1

ISSN 2229-5518

Issues and challenges of Research: Originality Based Document Classification and Information Extraction

Hussam Eddin Alfitouri Elgatait& Wan Mohd Fauzy Centre for Instructional Technology and Multimedia Universiti Sains Malaysia

Penang, Malaysia hussamelgatait@hotmail.com, fauzy@usm.my

ABSTRACT— AS THE NUMBER OF PUBLICATIONS IN THE KNOW LEDGE DOMAIN IS STILL INCREASING, NATURAL LANGUAGE PROCESSING (NLP) AND TECHNIQUES FOR INFORMATION EXTRACTION FROM TEXTS BODY HAVE BEEN APPLIED WIDELY IN DIFFERENT FIELDS. IN ADDITION, THE SCIENTIFIC WRITING REQUIRES ARGUMENTS AND EVIDENCE, THE RESEARCHERS MUST BE CERTAIN THAT THE SOURCE OF INFORMATION IS RELIABLE, GENUINE AND VALID IN ORDER TO CONVINCE THE READER AND ENRICH THE KNOW LEDGE. THEREFORE, THIS PAPER HIGHLIGHTS THE CURRENT ISSUES AND CHALLENGES OF USING DIFFERENT INFORMATION EXTRACTION TECH- NIQUES.

INDEX TERMS: INFORMATION EXTRACTION; SENTENCE CLASSIFICATION; COMPUTATIONAL LINGUISTIC; NATURAL LANGUAGE PROCESSING (NLP); KNOW LEDGE

DOMAIN.

1 INTRODUCTION

——————————  ——————————
echniques for extract information’s from texts body have become several and various. In addition to the above- mentioned, the targets of information extraction from texts have become urgent to decision-making; existing ap- proaches focused on extract structured information from un- structured and identify some features from a collection of documents (i.e. location, relations between entities, author name, date...etc). In the same context, Information Extraction (IE) systems appear as tools to assist the information access, by extracting the parts that suitable to fill in a set of pre-defined
output slots from the from a collection of documents [1]. Data extracted can be used to storage in database or used to presen- tation the data to user direct.
On the other hand, not all information available on the in- ternet reliable (i.e. wikipedia, Yahoo, forums, etc) and these sites providing the information for learners and researcher in anytime and anywhere by provide them with full accessing and availability 24 hours a day, and seven days a week. And because the scientific writing require arguments, evidence and the evidence, so, the researchers must be certain that the source of this information it is judgements scientifically and Original sources, to convince the reader and enrichments the knowledge. in the domain of computer assisted language learning and natural language processing and information extraction technique is an important aspect of digital docu- ment, with information extraction the researchers/students can analyse and develop their trends of topics and relationship between the documents.
When the individual extract these metadata they can get
high accuracy, but because we dealing with large number of these documents, progress of computer technology and in- formation extraction technique demonstrated the efficiency and accuracy in this the domain[2], metadata extraction has become more interesting and researchers focus, we will focus on this study on the metadata extraction is refers here to ex-
tract metadata information from the references (i.e. Author, title, volume, etc) in the document and the relationship be- tween these references in the candidate documents, (that ad- dress some problems such as errors in writing references and various styles of references) to proof the originality of the document and then prove the Information in Wikipedia it is reliability or not reliability, so, to is ensure that the informa- tion content matching with the content in the collection of documents, and then extract the metadata about this candi- date document are to understanding the concept of free-text documents[3], text-to-text semantic[4], pattern-matching tech- niques [5], and latent semantic analysis approaches similarity existing approaches [6].

2 ISSUES AND CHALLENGES

Internet can be a great tool for learning and researching in- formation, it's provides a useful information available to re- searchers and students. Ellsworth described the internet in- formation as "a worldwide personal library," (Ellsworth 1995). Retrieval these information easy and safe time and effort by using the internet and available online any anytime anywhere. In addition, this information available around the world, stu- dents can learn, proves to be a good choice for supporting re- search and develop their search abilities. This kind of learning can be very useful for students as well as teachers. But some of these websites on the internet (i.e. wikipedia, E-Answers on- line, forums, etc) is cooperative effort of tens of thousands people, anyone can add and modify its content, these informa- tion may not constitute an acceptable source for a research paper.
Some of these websites does not publish original papers, Dic-
tionary.com defined the Originality is "the quality or fact of being the product of individual creation that warrants copyright protec-

http://www.ijser.org

The research paper published by IJSER journal is about Issues and challenges of Research: Originality Based Document Classification and Information Extraction 2

ISSN 2229-5518

tion for a particular work regardless of novelty”. The "original pa- pers" refers to material (i.e. information, facts, ideas, etc) not all material published by reliable sources. This means not all material added to internet websites attributable to a reliable published source, so, must to find reliable sources about this topic, you should discover reliable source, and Because these websites is not the place to reliable information, and these materials that is challenged or likely to be challenged must be supported by a reliable source. Ma- terial for which no reliable source can be found is considered origi- nal research. Any cite you can show edit is not original research is to cite a reliable published source that contains the same material. Even with well-sourced material, if you use it out of context, or to advance a position not directly and explicitly supported by the source, you are engaging in original research.
Although the internet websites provided useful information for
user in many fields (i.e. medicine, science, toxicology, cancer re-
search and information drug, etc), and found that the depth these websites and the coverage was of a very high level, comparable in many cases, the coverage of databases, a doctor and is much better than Well-known reputation and national media. However, but still there many gaps when you use it, and some important issues about some information on products harmful and a source of great concern to areas such as medicine.
As in mentioned above, most of the people such as the re-
searchers and students are looking through the Web for the infor-
mation which are related to their research topics. The relative in-
formation of the research topics can be accessed from different re-
sources (i.e. E-Journals; books published by university presses or published by respected publishing houses; magazines and main- stream newspapers). The researchers are focusing on the rightness of the gained information in order to consider the source as a refer- ence. So many researchers are referring to the articles that can be accessed from the e-Journals (i.e. IEEE, ACM …) because these e- Journals have restrictions on the published articles. On the other hand, most of information resources do not have any restriction on their published articles such as; wikipedia, forums, etc. Conse- quently, invalid information may be published and then these re- courses can’t be considered as references. Based on these issues there is need to a method that can measure the information validity of these resources is in demand in order to be considered as a refer- ence, so, we need further investigate for prove the information in the internet if reliable, genuine and valid or not in order to convince the reader and enrich the knowledge.
A number of researchers [7-11] addressed the drawbacks of the current detection tools that are:
• Lacking to distinguish correctly cited text from plagiarized text.
• lacking to involve the books category.
• lacking to detect plagiarized words based ideas.
• Lacking to process textual images for similarity checks.
However, indicated the way that most of detection methods
follow for analysing analysis the document originality, which
shows that although these tools provide excellent service in detect-
ing matching text between documents, but it’s still lacking to dis-
tinguish correctly cited text from plagiarized text is one of the se- rious drawbacks of these tools. That is why utilize new technique is still needed.

3 MOTIVATION

As demonstrated in this document, the numbering for sections upper case Arabic numerals, then upper case Arabic numerals, separated by periods. Initial paragraphs after the section title are not indented. Only the initial, introductory paragraph has a drop cap.

4 LITERATURE REVIEW

Current trend detection methods fall into two categories: fully- automatic and semi-automatic. In fully-automatic approach, a list of emerging topics is developed afterwards researcher pursues these topics and the evidence to determine which truly emerging trends are. Semi-automatic approach on the other hand, requires user to input a topic then it provides the user with the evidence whether the input topic is truly emerging [12]. Most of the existing trend detection systems focus on keyword matching, statistical techniques and link analysis. Currently there are emerging efforts to employ Semantic Web technologies to provide enhance informa- tion search and retrieval mechanisms. The Semantic Web has been desired as an extension of the current Web, which makes a well- defined meaning of information to enable a better connection be- tween computers and people to work together [13]. Ontology is used to visualize knowledge on the Semantic Web. Recently, on- tology is a formal representation of a set of concepts within a do- main into a machine-readable format that is also understandable by humans, consisting of entities, attributes and relationships [14].
Figure 1: A generic structure of document source comparison based on plagiarism detection system [11]
Figure 1 presents the traditional method for comparing two or more documents and to reason about degree of similarity between them, it is needed to assign numeric value, so called, similarity score to each document. This score can be based on different me- trics. There are many parameters and aspects in the document which can be used as metrics.

http://www.ijser.org

The research paper published by IJSER journal is about Issues and challenges of Research: Originality Based Document Classification and Information Extraction 3

ISSN 2229-5518

According to Bahrami in [7] highlights the importance of the physical components and sub-components of the e-document to be converted to an XML e-document that can be used to match simi- larity or document originality. Bahrami indicated the main issues in linking the conversion process that justify the issue in identifying the components and sub-components of the e-document. Bahrami adopted a methodology, blocking scheme to identify semi- structured e-documents’ components and sub-components. The methodology was applied on 50 semi-structured e-documents of a specific type and the identification of their components and sub- components were achieved with 100% accuracy.
Deeper, Mahdavi et al. (2009) reports the prime issues in ana- lyzing and interpreting document contents. Trend detection in scientific publication retrieval systems helps scholars to find rele- vant, new and popular special areas by visualizing the trend of input topic. Authors aimed to describe the difficulties of previous researches to obtain a suitable classification model for research top- ics that it’s cited by other researchers. Mahdavi et al. developed a technique that combines both of semantic components and ontolo- gy classification to provide an advance functions for trend detec- tion in the context of scholarly Semantic Web system (SSWeb) [8].
Shaparenko and Joachims (2009) address the main issues in
specifying the cited articles that most of research papers referred to during their research and how text mining provide readers with automatic methods for quickly finding the key ideas in individual documents and whole corpora. Shaparenko and Joachims pro- posed a statistically well-founded method that aimed to determine and analyze the original ideas that a document contributes to a corpus, focusing on self-referential diachronic corpora such as re- search publications, blogs, email, and news articles. This model works on indicating the original content through a combination of impact and novelty, and it can be used to identify the most original passages in a document. Shaparenko and Joachims evaluated the developed model among synthetic and real data. The obtained result showed that the model outperforms a heuristic baseline me- thod [9].
Finally, Van et al. (2008) reports the importance of providing a
way for detecting that original author for a certain document, which consider to be an important role in specific domains. It’s also showing how the current methods in identifying and detecting authors work are lacking to describe and generalize the full fea- tures for this class of documents. Van et al. proposed a new ap- proach for detecting false documents using a document signature obtained from its intrinsic features: bounding boxes of connected components are used as a signature. The work process for this ap- proach lied on utilizing the optimal document alignment to build a model signature that can be used to compute the probability of a new document being an original one. Preliminary evaluation shows that the method is able to reliably detect faked documents.
The process starts with painting the original black pixels from
the document in blue; and then, the black pixels from the original
document will be painted in red. The procedure of this approach
will indicate the differences in the red and blue pixels. It can be
seen that the copying process seems to move blocks up and down: left part the copy is too far down, the middle parts quite well and the right part is again to far down [10].

5 CONCLUSION

This paper aimed to put the sight on the current challenges and issues faced by different academicals institutions towards using natural language processing for checking research originality by classifying and extracting information inside the research papers. Prior researches on the current techniques were addressed to verify these issues.

6 REFERENCES

[1] [1] J. Turmo, et al., "Adaptive information extraction," ACM Comput. Surv., vol. 38, p. 4, 2006.

[2] [2] Z. Ni and H. Xu, "Automatic Citation Metadata Extraction Using Hid-

den Markov Models," presented at the Proceedings of the 2009 First IEEE In- ternational Conference on Information Science and Engineering, 2009.

[3] [3] R. Williams and J. Nash, "Computer-Based Assessment: From Objective Tests to Automated Essay Grading. Now for Automated Essay Writing?," in Information Systems: Modeling, Development, and Integration. vol. 20, J. Yang, et al., Eds., ed: Springer Berlin Heidelberg, 2009, pp. 214-221.

[4] [4] R. Mihalcea, et al., "Corpus-based and knowledge-based measures of

text semantic similarity," presented at the Proceedings of the 21st national con- ference on Artificial intelligence - Volume 1, Boston, Massachusetts, 2006.

[5] [5] N. K. Nikitas, "Computer Assisted Assessment (CAA) of Free-Text: Literature Review and the Specification of an Alternative CAA System," 2010, pp. 116-118.

[6] [6] P. Dessus, "An Overview of LSA-Based Systems for Supporting Learn- ing and Teaching," presented at the Proceeding of the 2009 conference on Ar- tificial Intelligence in Education: Building Learning Systems that Care: From Knowledge Representation to Affective Modelling, 2009.

[7] [7] A. A. Bahrami, "A Blocking Scheme for Identification of Components and Sub-Components of Semi-Structured E-Documents," in Fifth Internation- al Conference on Information Technology: New Generations, NW-US, 2008, pp. 943-948.

[8] [8] F. Mahdavi, et al., "Semi-Automatic Trend Detection in Scholarly Repo- sitory Using Semantic Approach," World Academy of Science, Engineering and Technology, vol. 5, pp. 224-226, 2009.

[9] [9] B. Shaparenko and T. Joachims, "Identifying the original contribution of a document via language modeling," Machine Learning and Knowledge Dis- covery in Databases, pp. 350-365, 2009.

[10] [10] J. Van, et al., "Document signature using intrinsic features for counterfeit detection," Computational Forensics, pp. 47-57, 2008.

[11] [11] J. Jaya, "Plagiarism Detection Techniques," Cochin University Of Science

And Technology, 2007.

[12] [12] B. Aleman-Meza, et al., "Template based semantic similarity for security applications," Intelligence and Security Informatics, pp. 621-622, 2005.

[13] [13] K. Verma, et al., "Meteor-s wsdi: A scalable p2p infrastructure of regi- stries for semantic publication and discovery of web services," Information Technology and Management, vol. 6, pp. 17-39, 2005.

[14] [14] M. A. Ismail, et al., "Semantic support environment for research activi-

ty," Journal of US-CHINA Education Review, vol. 5, pp. 36-51, 2008.

http://www.ijser.org