Grant-in-Aid for Scientific Research (A)[English]

Grant-in-Aid for Scientific Research <KAKENHI>(A)

2022 〜2026

Elucidation of the Paper Road by data science.
-Based on Quantitative, Qualitative research and AI Multidimensional analysis-


Research representative

Koji Shibazaki (Aichi University of the Arts, Faculty of Arts)Research contributor


Naoki Kamiya (Aichi Prefectural University,Department of Information Science and Technology)

Tomohiro Banno(Aichi University of the Arts, Faculty of Arts)、

Mitsuko Honda (Aichi University of the Arts, Faculty of Arts)、

Akiko Iwata(Preservation and Restoration Institute,Researcher)

Tomoko Kenyama(Fujita Medical University, Shiga University until last year)


There are diverse paper cultures throughout the world. Paper has been an important medium in the development of civilization since prehistoric times. It continues to evolve and gradually diversify in accordance with the most advanced technology of its time and place, which was established with the emergence of human culture.

In the leftover fibers and components of paper, remnants of several paper cultures can be identified. This research focuses on the “Paper Road” prior to the 18th century, when papermaking technology became standardized, and clarifies the reality of paper distribution in a world where papermaking technology has been severed and cultural continuity is uncertain. The study is a collaborative multinational effort to pave the route for recovery.

Utilizing current portable microscopes and artificial intelligence for deep learning, this research collaborates with colleges, museums, and libraries in Asia, Europe, and the United States in the field of data science. Utilizing observation and photographic technology, a multifaceted image analysis system is employed to identify fiber structure and trace components. This is the first study initiative in the world to clarify new historical facts from an art and cultural history perspective.

(1) The academic context of this study, which represents the core academic “problem” of the research topic.

This study, which focuses on the “Paper Road” that solves many unanswered questions, tries to determine what paper culture existed in historical time series and is employed in data science. This research aims to unify the disciplines of paper culture, to uncover the truth concealed inside paper culture, and to investigate the diversity of paper culture.

This study employs current portable microscopes to acquire additional samples, employs deep learning for image analysis, and incorporates the multifaceted analysis results produced from multiple observation, measurement, and image simulation techniques. Through a simple and non-destructive method of photographing images, we intend to greatly enhance paper analysis.

This investigation was motivated by the Core-to-Core Program, which ran from 2017 to 2021. This program comprises in particular: 【The research for the culture of contemporary Hand-Made Paper and artistic expression. ~With the focus on the revival of Samarkand paper~】,Application Procedures for Fund for the Promotion of Joint International Research and 【The research on propagation of the world paper culture and elucidation of “Samarkand Paper”】.

It is derived from practice, analyzes the composition of paper using artificial intelligence, and combines many analysis systems to significantly enhance the method for elucidating paper entities.

From the perspective of practical study on paper archaeology and resurrection, this paper takes Eurasia as its research area, exploits paper resources that have survived the ages, and analyzes the composition and distribution of paper, which are mostly unknown in the modern world.

Academic credentials. The topics linked to “Paper and Artistic Representation” are summarized in Figure 1.

It is essential to have standardized papermaking techniques before the 18th century in order to decipher the paper culture of the world.

(ⅰ)From China to Japan, people in the East seek a variety of plant fibers as a raw resource. (Various plant skin fibers that can be used as papermaking raw materials have been researched.)

(ⅱ)Elucidation of components such as paper processing procedures and painting and calligraphy expression techniques for various uses. (Elucidation of elements such as the method of paper processing based on application and the manner of expression in calligraphy and painting.)

(ⅲ)In the West, investigate the migration of paper from China to Central Asia in the second half of the eighth century.

(ⅳ)Along with excellent paintings such as the Koran and gold paint, the painting skills and support of Miniature (meaning to the carrier used for calligraphy and painting, namely paper) replaced parchment in the Islamic world.

(ⅴ)In addition to the Samarkand paper, the origin of foreign paper, the ambiguous connectivity between modern papermaking, etc., are highlighted as significant issues.

From the perspective of creative practice, such as “paper and artistic expression,” the expression methods pursued by each era are distinct; consequently, diverse paper and processing techniques have arisen in response to the changing needs of the times.

Paper from around the globe collects a variety of cultural essences. Although the preservation and restoration of cultural assets has been subjected to rigorous repair and duplication, scholars have also performed similar research for paper and artistic expression. In accordance with the non-destructive and non-contact principles, research can still be conducted even if the type of paper cannot be accurately estimated.

There are numerous unknowns in the original theory on the spread of paper throughout Eurasia. Japanese washi and other Shosoin documents are among the few instances that can be dated back to 1300 years ago. With the rise and fall of nations throughout world history, paper culture will also flourish and perish. wiped and then revised.

Considering the history of research thus far, the following challenges exist in domestic and international research.

(ⅰ)After the papermaking technique was discontinued, study ceased, and cultural history research continued without the ability to examine the information contained on the paper.

(ⅱ)There are numerous unconfirmed facts concerning paper culture around the world, but in previous studies, only theory was used and a huge number of samples were not verified.

(ⅲ)Paper evolves with representation, and it is crucial to investigate the history of technology and culture from a scientific standpoint and to do thorough restoration work as an examination of artistic practice.

Consequently, this study integrates art and data science, employs cutting-edge technology, employs entity parsing methods on a worldwide scale, investigates the path of paper diffusion in practice, and solves paper analytics issues.


(2) The goal of this study and the academic originality and creativity

The objective of this study is to compare and differentiate the resemblance and identity of fiber data with other paper with high precision by shooting different papers from across the world with a portable microscope camera.

[First examination]

For paper of known age, we explore by visual observation and photography with a portable microscope camera, collect as many fiber images and paper-related data as possible, and develop a database for educational purposes. (Quantitative evaluation)

[Second examination]

In the second examination, we categorised and displayed these photos using deep learning to analyze the latent characteristics of the papers contained in the data.

[Third examination]

In the third examination, we gathered the original damaged portion of the characteristic fiber, or the naturally fallen paper fiber, and conducted a scientific analysis (JIS P8120, paper, paperboard). According to an optical microscope analysis of the C dyeing solution, the test technique for pulp fiber composition is based on the C dyeing solution. We study the fiber characteristics, combined components, etc., and identify the raw material and paper quality of the paper using the aforementioned test technique (the state and characteristics of the paper). (Quality evaluation)

Using techniques of deep learning, we iteratively connect this information with paper-based characteristic analysis data.

In addition, the [First examination] is quantitative, and a data upload mechanism has been developed to facilitate the collection of photos. As a way that can collect massive volumes of data from around the world, it provides a system that takes and uploads simple tiny photos on its own. After all, certain old texts are restricted to research contract schools, facilitators, and specialists, and some private collections cannot be lent out.

To date, the pre-processing for photos entered into deep learning consists of adjusting the image resolution, trimming the unfocused portion, and selecting a sharply focused portion of the patch. Deep learning leverages the EfficientNet architecture. Although this analytical procedure is in its experimental phase, it has achieved a recognition accuracy of more than 90 percent in distinguishing cotton from linen. This is significantly superior to human visual and tactile perception. If you increase the amount of picture materials, enhance the resolution of processed photos, enrich the learning data, and employ a more sophisticated architecture, you can anticipate more accurate judgments.

Until now, the verification has been used to identify raw paper (paper that has not been coated with reagents or other components), but as a method for identifying complicated paper such as liniment and pigments, it should be photographed by various high-definition machines. The outcome is an image generating simulation related with a simple microscopic image, as well as the development of a multifaceted analytical system.

Even if paper made from the same raw material has different forms, its quality must be determined by: 1 the difference in fiber type, 2 the degree of fiber compression, 3 the mixing of other fibers, 4 the verification of paper smoothness, 5 the verification of the liniment composition, and 6 the verification of the mixing of other substances.

Utilizing high-resolution, deep synthesis, 3D imaging, and radiation analysis of ultra-high-precision devices, such as ultra-fine microscopic optical microscopes (effective for 16) or hyperspectral camera, would significantly increase the correctness of their, (135, categorized based on the optical spectrum), electron microscope photography (46), X-ray fluorescence, XRF (56), etc., depending on the images of other machines to make analog photography, reflected light photography, etc.  As a consequence of the inspection, the analysis of the learning data has been refined, the discrimination accuracy has been vastly enhanced, and the varieties of paper that can be distinguished have expanded.

The usual JIS P8120 paper test was also utilized in the research. However, it is impossible to perform an examination without a tiny quantity of unnecessary paper scraps, notwithstanding the modest amount of damage. This research suggests that the created system can become a more effective discriminant system.


(3)The process of producing this study idea, the associated domestic and international research trends, and the focus of this research are discussed.

Konigil Meros workshop in Uzbekistan, which was financed by JICA and UNESCO for paper restoration, came to the conclusion that Samarqand paper was manufactured from the phloem of mulberry trees based on studies conducted to date. But this is a perspective based on stories, such as the Battle of Talas bringing phloem paper to Samarqand.

However, the first formal microscopic examination in the history of Karabacek (Arab paper / Joseph von Karabacek), first labeled as Samarqand paper, discovered linen from ancient textiles, confirming the prior result (The Samarqand paper is from the phloem of a mulberry tree) various interpretations。

In addition, we examined the condition and features of paper in ancient manuscripts from the eighth to tenth centuries and discovered that many of the raw materials of paper contain remnants of cloth (linen, cotton), but no paper created from phloem such as mulberry was discovered.

On the other hand, an important historical fact is that the area is adjacent to Xotan in western Turkestan, and that paper containing mulberry phloem fibers has been discovered in an archaeological site dated to the 9th century. Therefore, when Konigil Meros (Meros Paper Mill, Konigil) produced paper, it should have also fully referred to the declaration that the raw material of Samarqand paper is mulberry, although this is not the case.

In light of the present state of paper analysis, it was discovered through a re-examination of the restoration of Japanese national treasures that the paper was initially designated as mulberry paper, but the outcome was mixed paper. In addition, there were instances in the Uzbek examinations when current conifer pulp was identified in the paper of old documents and documents from a specified historical period (the date of writing is known).

Investigating the composition of paper is difficult. In accordance with the non-destructive and non-contact standards, only a small number of samples may be collected for supplemental reference, and national treasures and significant cultural relics will not be subjected to repeated investigations.

Looking at this study, there is a great deal of uncertainty and illogic throughout history. In addition, referring to the research cases of the remaining documents or manuscript paintings in museums and libraries around the globe, scholars have conducted stringent investigations on paper when preserving and restoring cultural heritage, but they only concentrate on the records recorded there. However, the research on the actual paper is disregarded (referring to the detection of modern conifer pulp in the paper of the document of a specific age).

Regarding the inquiry of paper, there are very few research cases and academic information to scientifically comprehend the paper itself, and there has been no breakthrough to elucidate the facts. Detailed examination of paper materials will be a highly efficient method of scientific investigation.

This research is an approach based on a survey of the quantity of basic macro photography, first referring to a huge quantity of paper and then employing AI to see the object (referring to paper) in question.

If we explore the innumerable unstudied papers in the world, both time and money will be wasted, and if we wish to investigate significant cultural features, even the extraction and interaction of fibers is impossible.

Inferences about the age and type of the papers can be made based on the similarities found in the huge number of photographs of papers, which can be compared to examinations of other cases.

There are numerous documents relating to state authority and religion among the inherited cultural legacy, such as antique paper, thus the date of production may be determined in many instances.

In addition, although it is impossible to clarify the culture that has vanished and the history that has been wiped, there is a strong likelihood that museums or individuals would conserve the excellent cultural heritage represented by paper and art, so preventing wars.

If the methodology is widely employed around the globe, it can transcend regional and temporal borders, unearth the buried history of the past from uncovered facts and incorrect historical interpretations, and become the first scientific research method in the world.

The research concept and research activity 2-1 have been published in [26.August.2019, “The Research on Paper Propagation in the World and Samarkand Paper”]. The paper-based study of the resemblance and identity of paper on a global scale has been regarded as an attempt that has not yet been made, and further advancements are anticipated.