نوع مقاله : علمی پژوهشی

نویسندگان

1 دانشجوی دکتری علم اطلاعات و دانش‌شناسی، واحد تهران شمال، دانشگاه آزاد اسلامی، ایران

2 استادیار گروه علم اطلاعات و دانش‌شناسی، واحد تهران شمال، دانشگاه آزاد اسلامی، ایران

3 دانشیار گروه علم اطلاعات و دانش‌شناسی، واحد تهران شمال، دانشگاه آزاد اسلامی، ایران

چکیده

هدف: نمایه‌سازی تصاویر بر اساس موتورهای جستجو در بازیابی تصاویر نمایه شده مبتنی برمتن و مبتنی بر محتوا با استفاده از تکنیک دلفی است.
روش‌شناسی: ازنظر هدف، کاربردی و نوع پژوهش با استفاده از تکنیک دلفی است. جامعه آماری شامل کلیه متخصصان شاغل در دفاتر روزنامه‌های سراسری کشور در شهر تهران بوده که به‌تمامی پنج موتور جستجوی موردمطالعه اشراف و تسلط کافی داشته‌اند. تعداد این متخصصان 16 نفر به‌عنوان نمونه در دسترس در حوزه موردمطالعه بوده است. برای گردآوری داده‌ها با روش اسنادی به استخراج گویه های پژوهش و تدوین پرسشنامه دلفی پرداخته شد. متخصصان با بیست سؤال برمبنای طیف پنج گزینه‌ای لیکرت در طی چهار مرحله به اجماع کلی رسیدند. نتایج آزمون ضریب توافقی کندال برای مشخص نمودن میزان هماهنگی و اتفاق‌نظر میان پاسخ متخصصان در هر دور جهت تطبیق و مقایسه گزارش داده شد. با تائید پرسشنامه در بخش کیفی روایی محتوا، ضریب روایی کیفی محتوای پرسشنامه بالاتر از 78/0 و شاخص روایی محتوا بالاتر از 79/0 گزارش شد. پایایی پرسشنامه نیز بر اساس ضریب آلفای کرونباخ برابر با 916/0 سنجیده شد.
یافته‌ها: نشان داد موتور جستجوی Google از میزان بازیابی تصاویر بیشتری بر اساس شاخصه‌های ارزیابی‌شان برخوردارمی باشد. میان موتورهای جستجوی موردمطالعه از دیدگاه متخصصان درزمینۀ بازیابی تصاویر بر اساس نمایه‌سازی مبتی برمتن تفاوت معناداری در سطح 05/0  وجود نداشته است. موتور جستجوی Yandex، از میزان بازیابی تصاویر بر اساس نمایه‌سازی مبتنی بر محتوای بیشتری در سطح 05/0 برخوردارمی باشد. همچنین موتور جستجوی Google به‌صورت معناداری در سطح 05/0 ازلحاظ بازیابی تصاویر بر اساس حوزه‌های موردپژوهش کارآمدترمی باشد.
نتیجه‌گیری: مشخص شد که موتورهای جستجو عمومی گوگل نسبت به موتورهای جستجوی دیگر (yahoo، bing،pinterest و yandex) عملکرد بهتری در بازیابی تصاویر دارند؛ همچنین جستجو گران تصاویر در وب می‌توانند در انتخاب موتور جستجوی متناسب با نیاز خود و طراحان داخلی برای طراحی بهتر تصمیم‌گیری کنند. علاوه بر آن، نیز این نتایج به حوزه‌های مشابه قابل‌تعمیم است و نیز طراحان موتورهای جستجو درمی‌یابند که برای بازیابی بهتر تصاویر از کدام روش نمایه‌سازی استفاده نمایند.

کلیدواژه‌ها

عنوان مقاله [English]

Search engine-based image indexing in retrieving text-based and content-based indexed images using the Delphi technique

نویسندگان [English]

  • soudabeh derakhshandeh 1
  • Fereshteh Sepehr 2
  • zahra abazari 3
  • neshaneh neshaneh 3

1 PhD Student in Information Science, Department of Information Science, North Tehran Branch, Islamic Azad University, Tehran, Iran

2 Assistant Professor, Department of Information Science, North Tehran Branch, Islamic Azad University, Tehran, Iran

3 Associate Professor, Department of Information Science, North Tehran Branch, Islamic Azad University, Tehran, Iran

چکیده [English]

Background and Objectives: Image indexing based on search engines in retrieving text-based and content-based indexed images using the Delphi technique.
Methodology: This research is applied in terms of classifying research according to the method of data collection (research design), using the Delphi technique, and in terms of classifying research according to the purpose. The statistical population of the qualitative stage of this research included all the specialists who worked in the offices of national newspapers in Tehran and had sufficient aristocracy and mastery in all five search engines studied. Most of them, depending on the type of activity and field of work, work with a maximum of one or two search engines and are aware. Therefore, only 16 specialists were selected as the available sample for Delphi panel members. In the present study, the experts reached a general consensus with twenty questions in four stages, which were indexed based on a range of five Likert options (from very weak to very good). Given the low Kendall coefficient in the fourth round, the low agreement of the panel members, and the significance of the third round, it is concluded that there was no increase in agreement in this round, and the polling process should be stopped. In this study, after reviewing the existing texts and sources by documentary method (library study), 150 specialized questions in the field were collected. After discussion and exchange of views with experts and professors in this field, 31 questions were approved for implementation, which with 3 main components separately [image search engine evaluation criteria (nineteen questions), image retrieval based on text-based indexing (six questions), And image retrieval was performed based on content-based indexing (six questions). In the meantime, Hamshahri newspaper, with its various publications as well as provincial special issues, magazines (My Land, 24, Children, Health, Youth, Stories, Knowledge, Clues, Story Books, Advertising Brochures, Exhibition Brochures, etc.) [In addition to the 5 selected image search engines, there were other very good image search engines in this field that even ranked very well on Alexa (the international website for ranking sites and blogs); However, due to the lack of use of this language and inefficiency in Iran and the lack of a specific audience, we have ignored their choice in this study]. The results of 9 experts showed that the questionnaire's relative validity coefficient of 32 items out of 40 items was higher than the critical coefficient value of 0.78. However, the relative validity of the eight-item content was less than the critical coefficient and was omitted. Therefore, in the relative coefficient index, the content validity of 32 questionnaire items was confirmed. Also, the content validity index of the other 31 questionnaire items was higher than the standard value of 0.79. As a result, 31 questionnaire items were approved in terms of two relative content validity coefficients. In addition, Cronbach's alpha coefficient for the reliability of the questionnaire with 31 items on 16 experts showed that it is equal to 0.916, which is a high coefficient.
Findings: Google's search engine showed a higher image retrieval rate based on their evaluation criteria. There was no significant difference between the studied search engines from the perspective of experts in the field of image retrieval based on text-based indexing at the level of P <0.05. Yandex search engine has a higher content indexing based on indexing based on more content at the level of P <0.05. Also, the Google search engine is significantly more efficient at the level of P \u003c0.05 in terms of retrieving images based on the areas under study. The results of the present study indicate that: 1. In the search engines, from the point of view of experts in the field of image retrieval based on text-based indexing, there are more or fewer differences, so the highest average, in this case, belonged to the Google search engine, and the lowest average belonged to the Pinterest search engine. 2. Experts in image retrieval based on content-based indexing see some differences between the search engines, so the Yandex search engine showed the highest average in this case, and the Yahoo search engine had the lowest average among the surveyed search engines. 3. Regarding image retrieval based on evaluation criteria, there are some differences between the average search engines from the perspective of experts. Hence, the average Google search engine is higher than the average of other search engines, while the Yandex search engine, in this case, is the lowest. Has had an average. 4. From the experts' point of view, there are differences between the studied search engines regarding the most efficient search engine in retrieving images based on the researched areas. The results show that Google's search engine has a much higher average than other studied search engines and the lowest average. Has been to the Pinterest search engine.
Discussion: Google's general search engines perform better than other search engines (Yahoo, Bing, Pinterest, and Yandex) in retrieving images; web image searchers can also choose the search engine that suits their needs and interior designers for better design. In addition, these results can be generalized to similar areas, and search engine designers will find out which indexing method to use to retrieve images better. In conclusion, it is suggested that to pay more attention to the indexing and retrieval of images in search engines; designers should consider such features based on the main components identified separately: A. Criteria for image search engines (add a special code for image copyright, apply all components of this research as a menu and submenu in image search engines, notify the owners of the email (images) when uploading images for support) B. Consider image retrieval based on content-based indexing (possibility to combine multiple colors simultaneously, searchable image edges to be able to draw shapes, simultaneous searchability of image content information). 

کلیدواژه‌ها [English]

  • Image retrieval
  • search engine
  • image indexing
  • text indexing
  • content indexing
Abbasi Dashtaki, N., & Cheshmeh Sohrabi, M. (2019). Google, Yahoo and Bing Search Engines' Performance in the Persian information retrieval: A Fuzzy and classical evaluation. Librarianship and Information Organization Studies, 30(2), 96-111. [In Persian]
Abbaspour, J. (2005). Indexing images of challenges and approaches. Library Magazine, 9 (44), 167-177. [In Persian]
Abolghasem Mosalman, T., Momeni, E., & Haji Zeinolabedini, M. (2016). The Rate of the Precision in the Audio Visual Retrieval Resource by K-Means Algorithm. Library and Information Science Research, 6(2), 321-337. [In Persian]
Baxter, G., & Anderson, D. (1995). Image indexing and retrieval: some problems and proposed solutions. New Library World, 96(6), 4-13.
Choi, Y., & Rasmussen, E. M. (2003). Searching for images: The analysis of users' queries for image retrieval in American history. Journal of the American Society for Information Science and Technology, 54(6), 498-511.
Chu, H. (2001). Research in image indexing and retrieval as reflected in the literature. Journal of the American Society for Information Science and Technology, 52(12), 1011-1018.
Dhingra, S., & Bansal, P. (2020). Experimental analogy of different texture feature extraction techniques in image retrieval systems. Multimedia Tools and Applications79(37), 27391-27406.
Diamant, E. (2007). Modeling human-like intelligent image processing: An information processing perspective and approach. Signal Processing: Image Communication, 22(6), 583-590
Ebrahimi, N. (2017). An introduction to illustrating children's books. Tehran: Roozbehan Publications. [In Persian]
Esfandiari Moghadam, A., & Bahari Mowaffaq, Z. (2009). Features of search in web search engines a list-based approach.287 review. Quarterly Journal of Information Science and Technology, 25(2), 265. [In Persian]
Fathian, M. (2013). Content retrieval of images based on machine learning through user interaction. University of Kurdistan, Faculty of Engineering, Master's Thesis in Computer Engineering. [In Persian]
Hassan, I., & Zhang, J. (2001). Image search engine feature analysis. Online information review, 25(2), 103-114.
Kidambi, P. (2010). Human-Computer Integrated Approach towards Content Based Image Retrieval, Wright State University.
Lakdashti, A. (2009). Indexing and retrieving image data based on content and visual semantics in the image database. Islamic Azad University, Tehran Science and Research Branch, PhD thesis in Computer Science. [In Persian]
Ozendi, M. (2010). Viewpoint Independent Image Classification and Retrieval, Ohio State University. Review of Image Search Engines (2013). http://tasi.ac.uk/resources/searchengines. (Accessed 28 April 2013).
Poor Sistani, P. (2011). Evaluation of effective factors in content indexing and retrieval based on content in JPEG intensive field. Payame Noor University, Master Thesis in Computer Engineering, Faculty of Computer and Information Technology. [In Persian]
Qasemi Aluri, M., & Abbasi Dashtaki, N. (2019). Investigating the performance of public search engines and super engines in retrieving information in the field of information science and their degree of overlap. Quarterly Journal of Information Management Science and Technology, 5(2), 91-118. [In Persian]
Rigi, T., Dayani, M. H., & Fattahi, R. (2019). Phenomenology of qualitative research methodology in information retrieval studies. Quarterly Journal of National Library and Information Studies, 30(2), 18-38. [In Persian]
Scherer, R. (2020). Image retrieval and classification in relational databases. Springer: Cham.
 Souri, F. (2013). Search engine evaluation in image retrieval based on text and content based indexing. Master Thesis. Islamic Azad University, Faculty of Humanities, Department of Library and Information Science. [In Persian]
Tang, Y. P., Shimizu, E., Dube, G. R., Rampon, C., Kerchner, G. A., Zhuo, M., ... & Tsien, J. Z. (1999). Genetic enhancement of learning and memory in mice. Nature401(6748), 63-69.
Vassilieva, N. S. (2009). Content-based image retrieval methods. Programming and Computer Software, 35(3), 158-180.