Все
 
 
 


ИИ, МО, агентное и имитационное моделирование для исторической науки: возможности и ограничения
 
Код статьиS000000000016816-7-1
DOI10.31696/S000000000016816-7
Тип публикации Статья
Статус публикации Опубликовано
Авторы
Нанетти Андреа  
Должность: Директор-основатель Лаборатории междисциплинарных точных и эмпирических исследований (LIBER), Основатель и главный редактор проекта «Engineering Historical Memory»
Аффилиация: Наньянский технологический университет (Сингапур)
Адрес: Сингапур
Название журналаDigital Orientalia
ВыпускТом 1 № 1
Страницы12-18
Аннотация

В свое время Фернан Бродель высказал убеждение, что историк завтрашнего дня должен будет создать общий с программистами язык, который позволит сберечь сокровища человеческого опыта с помощью информационных технологий. В современном мире с высокой степенью глобализации общества рискуют потерять традиционные средства передачи наследия, прежде чем они смогут овладеть современными методами, позволяющими сохранить свои самые важные знания и ценный человеческий опыт. Сегодня надежды возлагаются на развивающиеся технологии цифровой истории, которые должны ответить на эти вызовы. Однако цифровая история все еще находится в утробном состоянии, потому что ей нужны универсальные онтологии (культурно-зависимые онтологии) для сохранения и анализа своего знания. Исторические записи должны быть всесторонне разложены на однозначные поля, чтобы их можно было считать машиночитаемыми.

Наша исследовательская группа, которая экспериментирует с инженерией исторической памяти (EHM, www.engineeringhistoricalmemory.com), привносит в исторические науки опыт применения машиночитаемых наборов данных из вторичных и первоисточников с дальнейшей обработкой с помощью искусственного интеллекта, машинного обучения и агентного моделирования.

Ключевые словаисторическая наука, цифровая история, данные сокровищ человеческого опыта, культурно-зависимые онтологии, искусственный интеллект, машинное обучение, агентное и имитационное моделирование
Получено16.09.2021
Дата публикации16.10.2021
Кол-во символов16325
Цитировать  
100 руб.
При оформлении подписки на статью или выпуск пользователь получает возможность скачать PDF, оценить публикацию и связаться с автором. Для оформления подписки требуется авторизация.

Оператором распространения коммерческих препринтов является ООО «Интеграция: ОН»
1 From a human sciences perspective, the dreaming about a Robot Scientist like Adam — the one created in the United Kingdom for Genomics by an interdisciplinary research team of computer scientist and biologists to identify “genes encoding orphan enzymes in Saccharomyces cerevisiae” [1] — or powerful AI algorithms — like the one developed by DeepMind to predict with unprecedented precision the 3-D structure of nearly all the proteins made by the human body [2, p. 635] — pose a series of reflections. Indeed, automation for human sciences has its own peculiarities. However, the discourse on formalisation through ontology and language can be instructive because the creation of learning algorithms needs conceptual solutions. In 2015 the computer scientist Pedro Domingos justifies his use, with poetic license, of the term master algorithm in the Prologue to his homonymous book and explains it in the context of the advancement of learning.
2 If it exists, the Master Algorithm can drive all knowledge in the world-past, present, and future-from data. Inventing it would be one of the greatest advances in the history of science… The Master Algorithm is to machine learning what Standard Model is to particle physics or the Central Dogma to molecular biology: a unified theory that makes sense of everything we know to date, and lays the foundation for… future progress… What it requires is stepping back from the mathematical arcana to see the overarching pattern of learning phenomena… Once we have the conceptual solution, we can fill in the mathematical details… convinced that learning is the key to solving AI… nothing could have more impact that teaching computers to learn [3, pp. xviii–xix].
3 To approach this titanic task, a robust method — for the identification of key academic disciplines and the definition of their major contributions to a new interdisciplinary theory or science — is provided by the work published in September 1959 by John H. Holland as Willow Run Laboratories Technical Memorandum 2900-52-R. The work surveys how four academic disciplines contribute to the foundation of “Automata theory (presently more formally named, at M.I.T. and The University of Michigan, The Communication Sciences)”: logic and meta-mathematics provide the axiomatic method and decision procedures (formal system), effective definition, and computability (Turing Machines); the studies on electronic computers contribute programming and simulations; information theory (Shannon’s mathematical theory of communication) is used for measure of information and channel capacity; neurophysiology offers theories on the formation of concepts in the nervous system, and in particular on the recognition of macroscopic patterns by a sequential process [4].
4 Fernand Braudel’s persuasion that the historian of tomorrow, to be able to survive as a human scientist, must become able and build up a common language, which can be used by programmers across all human sciences to write their computer programs, is clear from a historian’s perspective. From a computation and machine learning perspective, one key question needs to be addressed and answered, before taking any action. Why should the historian oversee such a critical task? [see 5; 6].
5 Referring to Fernand Braudel’s 1969 call for unity, it seems that human sciences need to build up a grammar for the computational language that will allow the reloading of the treasure of human experiences through artificial intelligence. Big human sciences data are embedded and embodied in all sorts of artefacts and media that over time and across space have been created by humans to encode their knowledge and values, to make a better use of them in the present and transmit them to the future generations as well. In this way, artefacts and media are considered and treated as knowledge aggregators ante litteram. These big data can be named “THE data”, where THE stays for Treasure of Human Experiences, and easily identified as the most valuable component of what is generally called heritage (i.e., what may be inherited) [7, pp. 83–84, 89].
6 Societies have always used their heritage (i.e., the treasure of human experiences) to remain resilient and to express their cultural identities. Now, it seems that the rapid development of our society is endangering this spontaneous cultural process of adaptation to change. Part of this social adaptation is enabled by the human knowledge and values that we transmit from one generation to the next. In the past, this transmission is passed down through artefacts, oral traditions, social rituals, and cultural practices. But given the increasing pace of technological development of our times, many traditional modes of knowledge and value transmission have become obsolete or at risk of vanishing. New media and non-conventional communications have arisen, creating new possibilities for cultural expressions. However, we may not have mastered effective ways to harness these channels to transmit our cultural heritage, especially in the areas of transmitting knowledge and its associated values to enrich human experiences and augment the resilience of our society [see 8].

всего просмотров: 538

Оценка читателей: голосов 0 

1. King R.D. e.a. The Automation of Science // Science. 2009. 324. Pp. 85–98.
2. Callaway E. DeepMind’s AI Predicts Structures for a Vast Trove of Proteins // Nature. 595. 2021, 29 July.
3. Domingos P. The Master Algorithm. How the Quest for the Ultimate Learning Machine Will Remake Our World. New York: Basic Books, 2015.
4. Holland J. Survey of Automata Theory. Ann Arbor (MI): The University of Michigan Willow Run Laboratories, 1959; репринт: Vasbinder J.W., Gao H. (eds.) Selected Papers of John H. Holland. A Pioneer in Complexity Science. Singapore: World Scientific, 2018. Pp. 1–24.
5. Para Limes. http://www.paralimes.ntu.edu.sg  (дата обращения: 28.08.2021).
6. Nanetti A. Overcoming Linguistic and Cultural Obstacles in the Transcultural (Re)-Reading of Primary Sources and Secondary Literature for Afro-Eurasian Pre-Modern History (1205–1533) // Mukherjee, R. (ed.) Order/Disorder in Asia: Historical Perspectives. Kolkata (India): The Asiatic Society, 2021. Ch. 16.
7. Nanetti A., Simpson A. Sharing our Heritage to Shape our Future. How Effective Are Multi-User Sharing Platforms in Supporting Collaborative Visioning for the Future, and why is Heritage Centre-Stage? // SOTICS 2015. The Fifth International Conference on Social Media Technologies, Communication, and Informatics (Barcelona, Spain, 15–20 Nov 2015). Wilmington (DE): IARIA XPS Press, 2015. Pp. 82–90.
8. Nanetti A. Defining Heritage Science. A Consilience Pathway to Treasuring the Complexity of Inheritable Human Experience through Historical Method, AI and ML // Complexity 2021 = Chen S.-H., Alfarano S., Shen D. (eds.) Tales of Two Societies: On the Complexity of the Coevolution between the Physical Space and the Cyber Space. Article ID 4703820.
9. Eco U. From the Tree to the Labyrinth. Cambridge (MA) & London: Harvard University Press, 2014. (перевод: Эко У. От древа к лабиринту. Исторические исследования знака и интерпретации / Пер. с итал. О.А. Поповой-Пле. М.: Академический проект, 2016.)
10. Lechte J. Key Contemporary Concepts. From Abjection to Zeno’s Paradox. New York: SAGE Publications, 2003.
11. Nanetti A., Cheong S.A., Filippov M. Interactive Global Histories. For a new information environment to increase the understanding of historical processes // Proceedings of the International Conference on Culture and Computing 2013 (Ritsumeikan University, Kyoto, 16–18 Sep. 2013). Los Alamitos (CA): IEEE Computer Society, 2013. Pp. 104–110.
12. Nanetti A., Cheong S.A. Computational History: From Big Data to Big Simulations // Chen S. (ed.), Big Data in Computational Social Science and Humanities. Cham (Switzerland): Springer International Publishing AG, 2018. (Springer Series on Computational Social Sciences). Pp. 337–363.
13. Young T., Hazarika D., Poria S., Cambria E. Recent Trends in Deep Learning Based Natural Language Processing // IEEE Computational Intelligence Magazine 2018. 13/3. Pp. 55–75.
14. Selected Works of Ramon Llull (1232–1316), translated by Anthony Bonner. Princeton: Princeton University Press, 1985.
15. Die Philosophischen Schriften von Gottfried Wilhelm Leibniz, herausgegeben von Carl Immanuel Gerhardt. Bd. VII. Berlin: Weidmannsche Buchhandlung, 1890.
16. Oeuvres Philosophiques Latines et Françoises de feu Mr de Leibnitz, par Rud. Eric Raspe. Amsterdam & Leipzig: Jean Schreuder, 1765.

Система Orphus

Загрузка...
Вверх