OCR Project

What is the best way of preserving and sharing cultural heritage like books and pictures? During my studies I am met with the obstacle of not finding books that were needed for classes since they were old and nobody took the time or had the resources to make them available. I assume, this would be the case for many technologically underdeveloped countries and countries with few resources. Is it possible to do something about it with few resources, based on the voluntary work of a few enthusiasts? Thanks to DH at the University in Oslo I have obtained insight in the possibility of how something can be done about it.

This term paper will be created in two parts. The first part will focus on the transformation of the book from hardcover to digitalized e-book by scanning the physical copy and then using Optical Character Recognition (OCR) software to make the content of the book machine-readable. One of the most used programs for this type of work is ABBYY, but since it is proprietary software I will try to find a free option, possible Tesseract.

In the second part of the term paper I will walk the reader through the process of publishing the e-book and making it available for public use. This will be done by Creating an Wikisource account and learn about legal criteria and guidelines for publishing material on internet and on Wikisource. Secondly, I will make proper metadata for this book on Wikisource and make description on how it works.

The book that I will work on is a travelogue describing Norway “Letters from Norway” by Isidora Sekulich, an Serbian intellectual who traveled to Norway in 1912, and published her book in 1914. This book it is a unique view of early travels to Norway with a South-East European view. I chose the book because it does not exist in digital form yet and the copyright has expired.

I would like to write this term paper in a way that it can used as a guideline for creating other, similar projects.

Preliminary sources:

https://en.wikipedia.org/wiki/Optical_character_recognition  (accessed 24.04.2018)

https://en.wikipedia.org/wiki/Tesseract_(software) (accessed 24.04.2018)

https://www.youtube.com/watch?v=QLXEYtvNfP (accessed 24.04.2018)

https://www.youtube.com/watch?v=jO-1rztr4O0 (accessed 24.04.2018)

https://github.com/Bockwinkel/OCR  (accessed 24.04.2018)

AI and the interpretation industry


Zhang Bei

AI (artificial intelligence) came of age in 2016 with the hallmark of the victory of AlphaGo over human Go master. And now AI is actually penetrating and shaping people’s life and different industries, including interpretation and translation industry. There are already many media reports that AI will replace human interpreters and translators some day and those reports impact people in this industry.And Chinese and foreign media discourse show different tones and perspectives towards the advent of AI and its relationship with interpreting industry. Therefore this paper collects main media outlets’ reports about how AI influences interpreting industry and compares the differences between Chinese and foreign media’s focus on this topic. The time span of the media discourse collection is from January 2016 to January 2018. And this study is based on frame analysis, a news discourse analysis theory. Frame analysis theory has three core concepts: frame, framing and framing effect. And this corresponds to discourse, discourse construction and discourse reception, namely production and communication effect.This paper aims to understand how media content, production and communication effect differ between Chinese and foreign media outlets on AI’s impact on interpreting industry. To apply the frame analysis theory, categories and frames should be determined first. And the researcher uses Voyant Tools to analyze words frequency in the collected media discourse, which helps researcher identify categories and conclude frames for those news reports. Then the researcher justifies how the frames can be used to analyze research subjects and explains the disparity between Chinese and foreign media discourse on AI’s influence on the interpreting industry. It’s estimated that Chinese media tends to sensationalize and overestimate the future of interpreting industry under the impact of AI and foreign media uses more scientific facts and technological advancement of AI to strengthen their hypothesis of interpreting industry future.



Title page


  1. Introduction
  2. Frame analysis theory

1) Definition

2) Core concepts

  1. Methodology

1) Research question and constructs

2) Research subjects and instruments

3) Categorization and determination of frames

  1. Discussion
  2. Conclusion




Entman, Robert. (1993). Framing: Toward Clarification Of a Fractured Paradigm. Journal of    Communication, 43(4), 51- 58.

Konig, Thomas. (2004). Anti-Semitism as Free Speech: A Case Study, paper presented to session  PCR 13-Methods, Research, Concepts, IAMCR Annual Meeting, Porto Alergre, Brazil, July25- 30.

Konig, Thomas. (2004). FrameAnalysis:TheoreticalPreliminaries,retrievedNovember13,2012,


Goffman, Erving. (1974). Frame Analysis: An Essay on the organization of Experience. New York:       Harper & Row.

Pan,zhongdang & Kosieki,Gerald. (1993). Framing analysis: An approach to news discourse.         Political Communication, 10(1), 57-75.

陈阳. 框架分析:一个亟待澄清的理论概念 (Framework Analysis: A Theoretical Concept to Be         Clarified). 国际新闻界, 2007(4), 19-23.

李秋杨. (2015).“中国制造”国际形象的媒体话语研究 (Research on the Media Discourse of   “Made in China” International Image). SinologiaHispanica, 1, 33-46.

李海波, 郭建斌. (2013). 事实陈述vs. 道德评价:中国大陆报纸对“老人摔倒”报道的框架分析 (Statement of Facts vs. Ethical Evaluation: A Framework of Reports on “Old People’s    Falls” in Chinese Mainland Newspapers). 新闻与传播研究, 1, 51-66, 127.

廖星, 刘建平, Nicola Robinson, 谢雁鸣. (2014). 定性研究方法之框架分析法 (Framework    analysis of qualitative research methods). 中国中西医结  合杂志, 34(5), 622–626.

潘忠党. (2016). 架构分析:一个函需理论澄清的领域 (Architecture Analysis: A Domain That Needs Theoretical Clarification). 传播与社会学刊, 1, 17-46.

邵静. (2011). 媒介框架论:新闻传播中框架分析研究的现状、特点与走向 (Media Frame Theory: Status Quo, Characteristics and Trend of Framework Analysis in News         Communication). 中国传媒报告, 1,    80-90.

万小广. (2010). 论架构分析在新闻传播学研究中的应用 (On the Application of Architecture   Analysis in the Research of News Communication). 国际新闻界, 9,5-12.

万小广. (2010). 王石捐款事件报道的媒介框架分析 (Analysis of the Media Framework of  Wang Shi’s Contributions). 传播与社会学刊, 2010(12), 79-110.

19th Century American Indian Folk-Lore and Legends

North American Indian Folk-Lore and Legends, by Anonymous (1890)

Assignment 2: Find a corpus and analyze it using Voyant

by Løvetann Ripoll

I chose to look through a small corpus of Indigenous American literary texts in preparation for my thesis next year, focusing mainly on the gender representation in the selected text with a hypothesis that female characters were placed lower hierarchically in terms of representation.

At the Project Gutenberg website, the bookshelf “Native American” held the anonymously written collection, North American Folk-Tales and Legends, published in 1890 (Anonymous, “The Project Gutenberg EBook of Folk-Lore and Legends: North American Indian, by Anonymous.”)

Upon inputting the text (excluding the Project Gutenberg notes and “pg” for page) the most frequent words were: said, great, man, came and lodge (Sinclair and Rockwell). “Man” made the top five and prompted me to focus on the textual representation of gender in Folk-Tales. The results indicated to me that “men” and male gods, especially Manabozho, came up more often in the text than female characters. To verify this, I learned how to search for variants. “Man” was mentioned 146 times, “woman” 51, “men” 75, and “women” didn’t make it into the 40s. “Wife” was used 70 times, while the titles of “warrior” and “chief” came up 58 and 71 times respectively, titles presumably referring to men (for more on women chiefs and warriors, see “Sitting Bull, Crazy Horse, Geronimo, Tecumseh and Other Heroes of Native Resistance.”).


My research question became whether female figures were less prioritized in the text, and when present, were so most often as “wife,” a title that tying them to men. Conversely, then, the question included looking at whether the men in the text were represented more frequently in total, and whether they, when represented, were so in roles such as “chief” and “warrior” (this based on which words showed up in the 59 most frequently used words: see Sinclair and Rockwell.) Already at this point the research became problematic, as I will go further into below. For one, nouns such as “hunter” and “chief” may have referred to both men and women; for another, without doing close readings it was not viable to do named character searches, meaning that there might have been more named female characters than male, undermining the thesis of gender imbalanced representation.


It is important at this point that this hypothesis, that of women being less foregrounded than men in the text, was based on the reasonable assumption that the author, though anonymous, would most likely be a Anglo-male writing from a perspective. The stories he (if male) collected might be misinterpreting gender roles and hierarchy or even purposefully misrepresenting them to make the text more palatable for Anglo-patriarchal readers. Other issues of representation could be language barriers, and not in the least that North American Nations are not monocultural – as of October 2016, there were “566 federally recognized tribes” in the U.S. alone (see “List of Federal and State Recognized Tribes”). Generalizations about gender are challenging enough within a single nation, let alone when considering many hundreds (and from what I hear, some American Indians recognize over 600 tribes or nations).


All this to say: it is not an unlikely hypothesis that the filters of culture, language and gender, the end-product of Folk-Lore and Legends is likely to be guilty of misrepresentation Native American gender roles and more. The “prefatory note” of the collection praises the “Fantastic imagination, magnanimity, moral sentiment, tender feeling, and humour” of the “primitive” “Indians of North America,” (Anonymous). Here we have evidence of Anglo-stereotype of the “noble savage” (see Duane Champagne’s explanation of the derogatory nature of this in “Noble Savages and Noble Nations”). In conclusion, I do not expect the collection of folk-lore and legends that follows such a preface to be much more informed – that is, my kind of “informed,” which means intersectionally conscious of privilege and oppression between races, genders etc. – and although this initial experiment with Voyant cannot count as evidence, further research, I am confident, will support my hypothesis.





“About Open Library | Open Library.” Accessed March 11, 2018. https://openlibrary.org/help/faq/about#what.


Anonymous. “The Project Gutenberg EBook of Folk-Lore and Legends: North American Indian, by Anonymous.,” Original book 1890. Accessed March 11 2018 at: http://www.gutenberg.org/files/22072/22072-h/22072-h.htm.


Champagne, Duane. “Noble Savages and Noble Nations.” Indian Country Media Network (news website), January 15, 2014. Accessed March 11 2018 at:  https://indiancountrymedianetwork.com/history/events/noble-savages-and-noble-nations/.


“Getting Started | Voyant Tools Documentation.” Accessed March 11, 2018. http://docs.voyant-tools.org/start/.


“List of Federal and State Recognized Tribes.” Accessed April 1, 2018. http://www.ncsl.org/research/state-tribal-institute/list-of-federal-and-state-recognized-tribes.aspx.


“Project Gutenberg.” Project Gutenberg. Accessed March 11, 2018. http://www.gutenberg.org/. [Native American bookshelf accessed March 11, 2018 and located at: http://www.gutenberg.org/wiki/Native_America_(Bookshelf)]


Sinclair, Stéfan and Rockwell, Geoffrey. “Summary.” Voyant Tools. 2018. Web. 1 April 2018. <http.voyant-tools.org>. Accessed April 1, 2018 at: http://voyant-tools.org/?corpus=b2e5232387dc6b8b70a4e81f3fa5202c&panels=cirrus,reader,documentterms,summary,contexts


“Sitting Bull, Crazy Horse, Geronimo, Tecumseh and Other Heroes of Native Resistance.” Indian Country Media Network (news website), July 28, 2016. Accessed March 11, 2018 at: https://indiancountrymedianetwork.com/free-reports/sitting-bull-crazy-horse-geronimo-tecumseh-and-other-heroes-of-native-resistance/. [Female warriors]





How to use Voyant Tools ? Analysis of the text “Time for Outrage” by Stephane Hessel

Time for Outrage

This body of text that I chose for an analysis with Voyant Tools is from Stéphane Hessel’s book  “Time for Outrage.(page 1 to 4) Stéphane Hessel was a member of the Resistance to the Second World War, a diplomat and a writer. In 2010 he published his manifesto “Indignez vous” in which he encourages the younger generations to maintain a power of indignation.

This manifesto is engaged politically and socially.The use of Voyant Tools and in particular the word cloud allows a clearer understanding of the text. Indeed, we see several words that reveal the fundamental ideas of the text as the word ‘resistant’ that is at the heart of the speech of Stéphane Hessel, this word is the most frequently used word: Voyant tells us that it has been used more than 30 times in 4 pages. This allows us to draw a first conclusion, the use of a repeated word can anchor in the memory of the reader, the idea that the author wants to communicate. In a broader vision we also see the words ‘youth’, ‘money‘, and ‘inequality‘, ‘democracy‘ giving us a good idea of the topics that will be discussed.


The illustration of the frequencies shows us the thread of the author’s thought. Thus we can see that the word resistance appears at the beginning of the text blatantly, then that it is the word story that takes over in the last segments. Based on this graph we can therefore correlate the terms.

Nevertheless, we must keep a critical perspective on the use of Voyant. It can be seen that the tool considers words as “to” or “was” as important words in the corpus. Moreover, what is lacking is what software of this type cannot yet never give out is the possibility to understand the ambiguity of words that will not be captured by Voyant.

All the philosophical and positive dimensions of Hessel’s discourse are transcribed schematically and synthetically. The use of a tool like Voyant can allow the rapid understanding of a complex text like that of Hessel and allows us to grasp the main thesis of the author quickly. For all that, can one truly grasp the speech of a man through the machine? This is one of the questions that is raised in the digital humanities.


Merwan Slamani

Practical use of Cirrus in Voyant Tools

What is Cirrus and what purpose does it have in Voyant tools? It is best explained by the Voyant tools documentation on Voyant-tools.org: “Cirrus is a word cloud displaying the frequency of words appearing in a corpus. Words occurring more frequently appear larger”.

For the purpose of showing the practical use of Cirrus in Voyant tools I have chosen the literary master piece ‘Iliad’, supposedly written by the blind poet Homer. Since the first time I read (around the age of 9) it, I was amazed by the age of this work, the war between Greeks and Trojans and my favorite hero Achilles. Let’s see if Voyant Tools can give me some new view on the theme of this work.

After uploading the text of the ‘Iliad’ into Voyant Tools first thing that catches our eyes is Cirrus (one big word cloud). In the centre of the word cloud we can see one big ‘’â’. 

By pointing over this letter with the courser we can see the number 1124. This is the number of times that the letter ‘â’ appears in the text.. Since ‘â’ is not revelent for my interest we can remove it by clicking on the “Define options for this tool” and ad ‘â’ to the list of irrelevant words and letters (I will do same with “a”).

Now we can see that the most frequent words are “son” (369); “Greeks” (313); “spake” (247); “men” (229); “Hector” (199).

This was a really big surprise for me! My favorite hero Achilles was not in the cloud! Was not this book about him and his deeds? In the upper left corner I clicked on “Terms” determined to find “Achilles”. And there he was. After typing “Achilles” in the field below the list of words and there frequency of occurance in the text I saw that he ended up on the 42.

place, below my second favorite hero of the story “Ajax”. Sadly “Links” did not work on the word “Achilles” so I was not able to figure out with what words “Achilles” co-occurred most often so I could perhaps understand why my favorite hero was not mentioned as often as I remembered when I read the ‘Iliad‘ as a kid.

Voyant and the Cirrus made me think that this story was not about my favorite hero, but about the relationship between fathers and the sons. Maybe “Hector” is the real hero of my favorite novel.