Tag Archives: crowdsourcing

Uncategorized

Hashtag Functions in the Protests Across Brazil

Reviewed By: Melissa Balok, Edward Pantoja, Marie Ingram, Chloe Noland, Emma Weinberg

Link to article: http://sgo.sagepub.com/content/5/2/2158244015586000

Introduction
In the world of Web 2.0, tagging behavior is emerging as an important area of qualitative research for studies interested in learning more about language, community, cultural identity, and much more. In this article, Recuero et al (2015) perform a qualitative study on the use of hashtags and tagging behavior on Twitter in a specific political and regional area: the June 2013 political protests in Brazil. By comparing the localized tweets to “a theoretical background of the use of Twitter and hashtags in protests and the functions of language” (Recuero et al, 2015), the study was able to manifest not only specific context-based trends in tagging behavior, but also identify larger trends in virtual communication, personal incentive, and emotional versus recruiting behaviors.

The article begins by giving a history of the political climate and state of Brazil at the time the hash-tagging sample was occurring. The authors explain how tweets at this time were used as both a mobilization tool by protesters, as well as a way for the community to keep abreast of real-time occurrences at the rallies. This personalization of politics is further exemplified through a discourse on the effects of social media on people’s social movements, personal lives, and documentation/spreading of information during critical times. Delving into the function of language, which can be broken up into six main sections, the authors applied these linguistic classifications to the conversational and organizational quality of tweeting. Core research questions included: what are the types and communicative functions of hashtags used during protests; how do co-occurrences of hashtags depict different meanings and functions; and what are the trends in hash-tagging behavior of users as events unfold over time?

Method
Dealing with an overwhelming content-base, Recuero et al (2015) methodically analyzed a large dataset of tweets. 2,321,249 tweets were analyzed between June 13-20 in 2013. These dates were chosen because they consisted of the most Twitter activity during the protest. To effectively create and organize a large dataset, 35 keywords were tracked and inputted into the open source software yourTwapperkeeper to archive tweets that contained keywords. Researchers then attempted to classify the meaning of hashtags and their co-occurrences. Answers to these questions helped to create a context around the function of hashtags and how different co-occurrences could depict different meanings.

In order to objectively analyze the large dataset, Recuero et al (2015) used a coding procedure to categorize hashtags. Jakobson’s (1960) model of six main language functions was used to categorize hashtags according to their linguistic and communicative purposes. From this basic foundation for classification, hashtags which were found together within the same tweet were also classified. Due to the overlap of functions in a single tweet, a hierarchy was needed to establish and identify the dominant function of each tweet. The criteria used to determine the dominant function was to ask, “What is the purpose of this message?” Co-occurrences were also used with the previous criteria in order to categorize tweets. Lastly, the 500 most retweeted tweets were analyzed using similar mechanisms to create a context around the quantitative analysis.

Findings/Conclusion
Classification of the hashtags within the dataset were thus painstakingly paired to each of Jakobson’s (1960) six language functions. Contextual hashtags that frequently related to geographical location – where the event was happening – were classified as “referential”. Hashtags which indicated user emotion, thought and opinion, including protesters’ demands, were labeled as “expressive/emotive”. “Conative” hashtags were those that urged action and served to motivate other protestors, and “metalingual” hashtags, which referred to the content of the tweet.

In regards to the co-occurrences, the authors found that the most prominent types of hashtags that occurred together were conative-conative, encouraging action and strengthening the message through emphasis. Conative-referential hashtags were also preeminent, combining the call to mobilize with a physical location. Referential-referential hashtags helped to spread contextual information. Other co-occurrences of hashtags functioned to mobilize through opinions/demands, to contextualize the tweet in entirety, or to “sign” the tweet. Re-tweeted tweets were also analyzed, finding that most re-tweets were focused on the live events of the protests as they unfolded.

Overall, the results demonstrated tagging behavior during the protests in Brazil to have several functions: to call others to action, to align and coordinate protesters, to share information including metadata regarding content, and to express and support opinions.

Questions and Further Research
It would be interesting to see studies conducted with the same classification of tweets and hashtags that this study created, but with examination of different protests in different countries. A comparison of the results with the protests in Brazil could make for a better qualitative understanding of how different countries use hashtags, in addition to furthering examination of tagging behavior across countries. Additional questions that come to mind: do Internet users use hashtags in the same way, regardless of language and country of origin? This could lead to bigger behavior studies regarding humans and Internet behavior in general and how humans adapt to technology. Alternately, is there evidence of similar tagging behavior in applications that allow more than 140 characters per post? Recuero et al (2015) briefly mention that the character limit on Twitter could potentially cause users to eliminate tags that are not as important. This leads one to ask, what information about the protests is missing from Twitter? Could additional information be found on alternative ICT platforms, such as Facebook?

References:

Jakobson, R. & Sebeok, T. (1960). Linguistics and poetics. Style in Language, p. 350-377. Cambridge, MA: MIT Press.

Recuero, R., Zago, G., Bastos, M.T., & Araujo, R. (2015). Hashtag functions in the protests across Brazil. Sage Open Journals, published 11 May 2015, doi: 10.1177/2158244015586000

Uncategorized

Using Open Source Tools to Create a Mobile Optimized, Crowdsourcing Translation Tool

By Molly Sherman, Jared Fair, Nicole Josephson, Dee Ann Huihui

Link to article: http://journal.code4lib.org/articles/9496

Article synopsis and core research question(s):
One of the main questions Oregon State University Libraries and Press (OSULP) described in this article was the need for a cost-effective solution to publish books in local languages and dialects using crowdsourcing. OSULP intends to create a tool that will benefit children ages two to five years old, while still allowing older children to benefit as well. OSULP found that children are taught to read later in their lives and in some cases, such as in Busia, are taught Swahili up until about third grade and then are taught in English from then on. With so many dialects, it is difficult to find a tool allowing for many dialects to be available.

This article discusses how to find or create a tool allowing for users, initially in Africa, to vote on the correct dialect of children’s books in order to make them available. With creating a tool such as this there are many aspects that must be considered such as the tools simplicity of its interface and usability, is there the ability to access it offline, what are the screen limitations, how quick can users vote, can users edit the books’ pages to add a new translation, and also the need for a system that takes the highest voted sections and combines them to create a completely translated text. OSULP created their tool in GitLab in order to keep the tool and its beginnings open only to their staff; hoping to later release it through GitHub for others to tailor it to their own needs.

Methods used to answer research questions:
In addressing how to create a tool that allows multilanguage translation in promotion of literacy for children in Africa, the following methods and applications of practice were used. First a network of libraries called Maria’s Libraries came together with OSULP to discuss technologies for consideration in a resource poor environment. It was decided that the testing methods would be delivered by mobile device interactions through a website using the following open source libraries: Wink Toolkit and Globalize3. Language translations were accessed as multiple database entries into an application called Ruby on Rails.

In creating this crowdsourcing platform the goal was to have a gateway tool for enabling users to translate folk tales and existing children’s books into their own languages and dialects. This practice was implemented with a simple interface offering a choice of languages. A username and password approach for users along with icons symbols and a voting process were part of a simple user setup. In order to support the most accurate translations of single languages and dialects, user abilities were set to allow for new dialect identification.
The following items were also included in the method design:
· Tablet Computers for easy interactivity.
· A simple interface utilizing language choices.
· Display carousel of books marked for translation.
· Easy navigation between book pages.
· A User voting process for each translation.
· A User editing process with original translation comparisons.
As part of the methods used, a voter incentive process based on comparison of translations was set as default. The simultaneously experience for users is to see all translations that have been made and be able to make the most accurate choice.

Findings:
As the article assigned to us; Using Open Source Tools to Create a Mobile Optimized, Crowdsourced Translation Tool, was not written within a standard format for academic literature and captured only the first phase of an ongoing project, the findings were less formal and more linear than might be expected.
That said, the project explored in this article found that there were three vital parts that presented significant problems and needed to be addressed: administration; additional crowdsourcing and offline functionality.
The administration woes were two fold: 1) the participants were unable to use the interface to upload stories that contained both images and texts, and, 2) the administrators had no ability to override the participants intentions and this fault left them open to “poisoned” or unreliable records.
Additionally, without further crowdsourcing, many users were unable to recommend new languages and dialects for stories or books to be translated into, rendering some of the work inaccessible.
Finally, authors also suggested that without improved offline capabilities the stories and participants work would be greatly diminished.

Conclusion:
The broad, main intent of this project began as an attempt to help literacy in Africa. They discovered that their tool and research were relevant beyond this intent. Indeed, it held exciting potential to do much more; “We are very excited about the possibilities of the usefulness of the platform as a way of publishing books in lesser-known languages or in regions where dialectic publishing is cost prohibitive.”

Unanswered questions you have and what future research might address:
After reading this article, there are a few questions that could be addressed. There is the potential to build a database of sub-dialects with this technology. Will the text be available for export after local translation has taken place? If enough members of a community participate in the translation process and if the texts could be exported into a central database, the potential for published works in each sub-dialect increases.

Will the translations move across the geography in a wave, so each new area has the closest possible translation available from neighboring villages? Or will each linguistic geographical area begin with the same base text? The linguistic preferences of each user will be remembered and set as the default when they log in, but once a book is translated into a specific dialect, will the rest of the tablets in the same geographic area automatically set their defaults to that dialect? Can a mother and child share default settings, while maintaining their own accounts?

What are the consequences of exposing sub-dialect communities to the written text of other areas? Could this tool be used to encourage the building of a shared national language? Are there possible negative consequences of this action, such as the diminishment of local dialects? Are there cultural and ethical questions that must be asked before proceeding? Future research could address the cultural shifts that will take place as a result of introducing written texts for early readers in local dialects to areas that have not previously experienced it.

A thoughtful attempt to answer your own questions:
The authors’ stated purpose of fostering early literacy in order to prepare children for school success raises the question of how this tool will actually support students entering classes that are not taught in their local language. One possible solution to investigate may be designing the product as a bilingual tool that displays the stories with text in multiple languages simultaneously — for example, in both the learner’s local language and the official language. In addition, the anticipated use of this product in language studies programs leads to the question of how it may be used to facilitate aural learning. In this respect, further development and enhancement to consider may be the integration of programming that allows contributors to upload audio recordings of translations along with text. And finally, considering the product’s potential as a tool to increase literacy, aid in language learning, and enable wider access to multicultural children’s literature, additional research and development is justifiable. Few things are free, however, and one must ask how future R&D is to be funded. Using the tool itself as an example, perhaps the answer lies in crowdsourced funding.

Uncategorized

Using Open Source Tools to Create a Mobile Optimized, Crowdsourcing Translation Tool

By Kira Painchaud, Molly Sherman, Jared Fair, Nicole Josephsen, and Dee Ann Huihui

Link to article: http://journal.code4lib.org/articles/9496

Article synopsis and core research question(s):
One of the main questions Oregon State University Libraries and Press (OSULP) described in this article was the need for a cost-effective solution to publish books in local languages and dialects using crowdsourcing. OSULP intends to create a tool that will benefit children ages two to five years old, while still allowing older children to benefit as well. OSULP found that children are taught to read later in their lives and in some cases, such as in Busia, are taught Swahili up until about third grade and then are taught in English from then on. With so many dialects, it is difficult to find a tool allowing for many dialects to be available.

This article discusses how to find or create a tool allowing for users, initially in Africa, to vote on the correct dialect of children’s books in order to make them available. With creating a tool such as this there are many aspects that must be considered such as the tools simplicity of its interface and usability, is there the ability to access it offline, what are the screen limitations, how quick can users vote, can users edit the books’ pages to add a new translation, and also the need for a system that takes the highest voted sections and combines them to create a completely translated text. OSULP created their tool in GitLab in order to keep the tool and its beginnings open only to their staff; hoping to later release it through GitHub for others to tailor it to their own needs.

Methods used to answer research questions:
In addressing how to create a tool that allows multilanguage translation in promotion of literacy for children in Africa, the following methods and applications of practice were used. First a network of libraries called Maria’s Libraries came together with OSULP to discuss technologies for consideration in a resource poor environment. It was decided that the testing methods would be delivered by mobile device interactions through a website using the following open source libraries: Wink Toolkit and Globalize3. Language translations were accessed as multiple database entries into an application called Ruby on Rails.

In creating this crowdsourcing platform the goal was to have a gateway tool for enabling users to translate folk tales and existing children’s books into their own languages and dialects. This practice was implemented with a simple interface offering a choice of languages. A username and password approach for users along with icons symbols and a voting process were part of a simple user setup. In order to support the most accurate translations of single languages and dialects, user abilities were set to allow for new dialect identification.
The following items were also included in the method design:
· Tablet Computers for easy interactivity.
· A simple interface utilizing language choices.
· Display carousel of books marked for translation.
· Easy navigation between book pages.
· A User voting process for each translation.
· A User editing process with original translation comparisons.
As part of the methods used, a voter incentive process based on comparison of translations was set as default. The simultaneously experience for users is to see all translations that have been made and be able to make the most accurate choice.

Findings:
As the article assigned to us; Using Open Source Tools to Create a Mobile Optimized, Crowdsourced Translation Tool, was not written within a standard format for academic literature and captured only the first phase of an ongoing project, the findings were less formal and more linear than might be expected.
That said, the project explored in this article found that there were three vital parts that presented significant problems and needed to be addressed: administration; additional crowdsourcing and offline functionality.
The administration woes were two fold: 1) the participants were unable to use the interface to upload stories that contained both images and texts, and, 2) the administrators had no ability to override the participants intentions and this fault left them open to “poisoned” or unreliable records.
Additionally, without further crowdsourcing, many users were unable to recommend new languages and dialects for stories or books to be translated into, rendering some of the work inaccessible.
Finally, authors also suggested that without improved offline capabilities the stories and participants work would be greatly diminished.

Conclusion:
The broad, main intent of this project began as an attempt to help literacy in Africa. They discovered that their tool and research were relevant beyond this intent. Indeed, it held exciting potential to do much more; “We are very excited about the possibilities of the usefulness of the platform as a way of publishing books in lesser-known languages or in regions where dialectic publishing is cost prohibitive.”

Unanswered questions you have and what future research might address:
After reading this article, there are a few questions that could be addressed. There is the potential to build a database of sub-dialects with this technology. Will the text be available for export after local translation has taken place? If enough members of a community participate in the translation process and if the texts could be exported into a central database, the potential for published works in each sub-dialect increases.

Will the translations move across the geography in a wave, so each new area has the closest possible translation available from neighboring villages? Or will each linguistic geographical area begin with the same base text? The linguistic preferences of each user will be remembered and set as the default when they log in, but once a book is translated into a specific dialect, will the rest of the tablets in the same geographic area automatically set their defaults to that dialect? Can a mother and child share default settings, while maintaining their own accounts?

What are the consequences of exposing sub-dialect communities to the written text of other areas? Could this tool be used to encourage the building of a shared national language? Are there possible negative consequences of this action, such as the diminishment of local dialects? Are there cultural and ethical questions that must be asked before proceeding? Future research could address the cultural shifts that will take place as a result of introducing written texts for early readers in local dialects to areas that have not previously experienced it.

A thoughtful attempt to answer your own questions:
The authors’ stated purpose of fostering early literacy in order to prepare children for school success raises the question of how this tool will actually support students entering classes that are not taught in their local language. One possible solution to investigate may be designing the product as a bilingual tool that displays the stories with text in multiple languages simultaneously — for example, in both the learner’s local language and the official language. In addition, the anticipated use of this product in language studies programs leads to the question of how it may be used to facilitate aural learning. In this respect, further development and enhancement to consider may be the integration of programming that allows contributors to upload audio recordings of translations along with text. And finally, considering the product’s potential as a tool to increase literacy, aid in language learning, and enable wider access to multicultural children’s literature, additional research and development is justifiable. Few things are free, however, and one must ask how future R&D is to be funded. Using the tool itself as an example, perhaps the answer lies in crowdsourced funding.