Grants:APG/Proposals/2018-2019 round 2/Wikimedia Indonesia/Impact report form
Metrics and results overviewEdit
The pandemic affected our program result and metrics especially for the newly register users, active collaborations, and volunteer hours.
|1. number of total participants||4,137||4,170||
|2. number of newly registered users||2,647||2,584||
|3. number of content pages created or improved, across all Wikimedia projects||32,430||35,718||
|4. Active collaborations||66||54||
|5. Volunteer hours||26,770||22,189||
Telling your program stories – all programsEdit
APG19_Impact_report#Community Building and SupportCommunity Building and SupportWe support the community, increase their capacity building and provide internet/books grants to support their contribution.
APG19_Impact_report#EducationEducationWe train new editors, capacity building for our volunteer trainers, and encourage education institution to involve more in Wikimedia projects.
APG19_Impact_report#GLAM & PartnershipsGLAMWe collaborate with various GLAM institutions to provide free local contents, from old magazines, coins and paper money, photo collections and cultural hackathon.
APG19_Impact_report#Content EnrichmentContent EnrichmentWe held cultural and natural photo competitions, writing competition, improve Wikidata contents, and build Javanese OCR for Wikisource.
Community Building and SupportEdit
We find several funded programs interesting to share to the world. Here are eight funded programs that were run by communities in Indonesia.
- A funded program named WikiSurau was set to document historical mosques (surau) in 10 regencies and cities of West Sumatra. Two volunteers carried out the programs by motorcycles for eleven days. Both integrated Wikidata, Wikipedia, and Wikimedia Commons in their planning for the program. As a result, there are 786 photos for 102 mosques captured. The photos complete the information on Wikidata and Wikipedia, and also add more online visibility to the historical buildings.
- Another program that added visibility to historical buildings was Wiki Istana Kalbar. This program photographed 11 palaces (istana) and other landmarks (mosques, traditional houses, museums) across West Kalimantan. It took 12 days for the two volunteers riding motorcycles from town to town to finish the program. In total, there are 249 photos about the historical buildings and landmarks published on Wikimedia Commons.
- Some programs focused on documenting local and traditional foods. One of them was WikiSanganan. Carried out by the Denpasar community in Bali, the program collected photos of Balinese cakes (sanganan) and created articles about them in the Balinese Wikipedia. Each cake was photographed from two angles. In total, there are 102 photos of 50 Balinese cakes and 50 created articles that use the photographs.
- WikiSalero was another program that documented foods. This program photographed foods of West Sumatra, varying from cakes, dishes, to beverages. Exploring 4 cities and 1 regency, the program resulted in 354 photographs of 119 kinds of food and beverage.
- A program named WikiBeksa recorded Javanese dance movements, especially movements from dances of the Surakarta Sunanate. In total, there are 15 movements video-recorded and 107 photographs about them. The program must be halted halfway due to pandemic.
- Another recording program was WikiWaditra. It video- and audio-recorded the Sundanese gamelan (local musical instruments). It recorded how to play the instruments and how they sounded in different pitches. Moreover, it photographed the instruments. Collaborating with Institute of Indonesian Arts and Culture, Bandung, it resulted in a total of 31 photos of musical instruments, 51 videos and 51 audios of how to play the instruments in two different pitches.
- WikiDialek was a program that recorded the Minangkabau dialects in West Sumatera. In collaboration with WikiTongues, it highlighted the differences between dialects of the Minangkabau language. In total, there are 10 videos highlighting lexical differences uploaded to Wikimedia Commons and 10 videos uploaded to WikiTongues’ YouTube channel under CC BY-NC.
- The last program was a creation of a 3.5 minute animation about the introduction of the Minangkabau Wikipedia. The animation was planned to air in WikiNusantara 2020 in March, later cancelled due to pandemic. Although the event was cancelled, the animation continued to finish and was uploaded to Wikimedia Commons.
Upon the pandemic declaration by the UN and the in-person activity prohibition by the Wikimedia Foundation, Wikimedia Indonesia followed by adapting the programs to online activities. In the community, all monthly community meetings were only allowed online. Social events were no longer allowed. On the grant, all grant funded programs must stop immediately since the grant required in-person activities. However, they can still continue online activities such as editing articles and uploading files to Commons. The grant scheme was changed from funding in-person programs to funding internet packages and book purchases for volunteers since both were possible to do from home. The new grant scheme was started in April and more than 30 volunteers had participated. On the conference, the committee stopped the 2020 conference and plans to hold it in 2021 if the situation improves.
Community-wise, Wikimedia Indonesia (WMID) launched three programs for APG19: (1) Community Building, (2) Community Support (Grant), and (3) Conference.
- Community Building (Pengembangan Komunitas) funded monthly meetings and a social event for each local community. So far, there are five active local communities (city-based) that we support: Bandung, Denpasar, Jakarta, Padang, and Yogyakarta. Until the end of APG19, most communities showed activities through monthly meetings, but only two communities were able to hold social events. Since all in-person activities have been prohibited due to pandemic, most monthly meetings can still continue through online platforms. Only the Jakarta community turned inactive for 4 months due to its internal problem. Two communities that were able to hold social events were Yogyakarta and Bandung communities just because they realized the events before March. The Denpasar community had to cancel their social event planned to happen in April immediately after WMID’s COVID-19 announcement was released. Other communities had not planned social events yet.
- Community Support (Grant) funded individual and community-based programs. There are 16 funding requests and. since COVID-19 warning was announced, even though all of them had been assessed as possible to fund. Among the funded programs, most of them are taking-photo activities, including photos of local foods, structures, musical instruments, and dances. Moreover, there are also book purchases for the local community’s source of references. A small portion of the activities are video recording of dance movements and playing musical instruments. Most of the funded programs had completed their in-person activities prior to pandemic warning; only one of the funded programs had to cancel and return the funds because it planned to run in-person activities in the second half of March, which was not possible considering the COVID-19.
- The second national conference WikiNusantara was planned to hold in March in Padang, West Sumatra. Since October 2019, a conference committee has been formed. A scholarship opening was announced throughout December 2019 and earlier this year, there were 55 scholarship recipients for WikiNusantara 2020. Learning from previous WikiNusantara in Yogyakarta, there were no significant problems related to this second conference. Until the first half of March, everything was ready; accommodations had been reserved, all travel tickets had been issued, performers had been paid, and programs had been established. The conference unfortunately had to cancel due to pandemic. At the end, the committee decided the conference to be held next year in Padang if the situation permits.
Lessons we learned are divided into two main categories: pre-pandemic and pandemic.
On community building, we should set standards for communities that cover not only monthly meetings and social events. Communities start to expand their influence through social media and instant messenger, hence more standards on online platforms. Moreover, there are possibilities that new local communities will form and a community may turn inactive, thus to strictly establish standards on community formation and status.
On community support (grant), there can be quicker steps for grant requests to get funded. The last grant funding could take more than a month on average for a request to finally get funded. Standards on who can propose and what can be proposed as programs need to establish.
Some activities related to community building and support can be rendered online during the pandemic such as monthly meetings and grant funding. The grant could be adapted and limited to fund internet packages and book purchases.
On the conference, it is not possible to render the event online since nearly everything has been paid. It might be possible to turn the conference online if there are still four to five months ahead to plan. Some nice ideas can be learned from online ItWikiCon and Wikimedia CEE Online Meeting. However, the committee also has to consider the Indonesian internet facilities and people’s culture. Only people living in Indonesian big cities may have good internet access. In addition, it may take a longer time for the Indonesian people to get familiar with online conferences.
The pandemic really changes our perspective on how to deal with local communities and their activities. Problems arose since some may find difficulties in accessing the internet or prioritize their internet package for other things. Before the pandemic, community members bond well when they are involved in offline activities. The pandemic turns all-online activities interesting at first, but boring these days. The current situation has to be evaluated so that the all-online activities can still be interesting to them.
Since 2016, we rebuild and maintain regular education program within Indonesia. We brought a lot of local contents, including Indonesian arts, culture, woman artist, and various Indonesian related topics. As we received enormous request to hold an education program from various partners, communities, education institution, and several other organizations, we did our first Training of Trainers Program for our volunteers. We trained them by building their capacity, including planning an event, how to make a good presentation, how to be a good presenter, also tweaking minor technical problem during the events. We hope they can be our “front guard” as we understand that we need their help to collaborate and to increase the visibility of Wikimedia movements and projects by training and educating various new partners and new volunteers, and they are precious assets to us because we are nothing without them.
Last year, our WikiLatih program reached 22 out of 34 provinces in Indonesia, with 10 Wikimedia projects. By cooperation with Lontar Foundation, this is the first time we went to Papua, our easternmost province. We got the chance to partner with Cenderawasih University to introduce and educate Papuan students with Wikipedia and open educational resources. Besides Papua, in this year’s WikiLatih, we also held some workshops in some provinces for the first time, they are West Kalimantan, East Kalimantan, North Maluku, South Sumatra, Riau, Jambi, Central Sulawesi, and Southeast Sulawesi.
We kept trying to organize Wikipedia editing workshops in local languages. We have even tested out to conduct Wikipedia editing workshops in bilingual, which involved editing in Indonesian Wikipedia, and editing in local languages Wikipedia, for example in Medan (involved Indonesian and Mandailing languages), in Gorontalo (involved Indonesian and Gorontalo languages), and in Banjarmasin (involved Indonesian and Banjar languages). These kinds of workshops were apparently proven effective in developing the local language communities because some participants were actually keen to develop Wikipedia in their local languages.
These workshops started by introducing them to the interface of Wikipedia through editing in Indonesian Wikipedia first. After the participants got familiar enough with the interface and understood the rules and policy of Wikipedia, they were required to write in their local languages Wikipedia.
Wiki Masuk Sekolah (Wikipedia Goes to School)Edit
Our Wiki Masuk Sekolah (WMS) program this year focused on three languages, they are Minangkabau, Sundanese, and Javanese. Therefore, we involved more communities in Bandung, Yogyakarta, and Padang in last year’s Wiki Masuk Sekolah program. The volunteers from those communities were responsible to carry out the program in our collaborated educational institutions in each region.
Last year, we have collaborated and signed MoU’s with 3 universities in the first semester, they are Andalas University (UNAND) in Padang, West Sumatra, Yogyakarta State University (UNY), and also Universitas Gadjah Mada (UGM) that are both located in Yogyakarta. In these universities, Wiki Masuk Sekolah became a part of one of their subjects. Apart from collaborating with universities, we also collaborated with some high schools, they are SMA 6 Padang (Padang 6 Senior High School) in Padang, West Sumatra and SMA 1 Cisarua (Cisarua 1 High School) in Cisarua, West Java.
While in the second semester, we started the 2020 with new collaborations. There were SMAN 9 Bandung (Bandung 9 High School) in Bandung, Sebelas Maret University (UNS) in Surakarta, and SMAN 5 Padang (Padang 5 High School) in Padang. However, this semester’s collaborations were postponed as well as the activities due to the coronavirus outbreak around the world. Even so, we have stated to the collaborated institutions that we will continue the program and pick up where we left when the situation gets better, and we are able to do offline activities again.
During last year of our WMS, one of our volunteers, who happened to be a lecturer in University of Gorontalo stated that he was interested to organize WMS and integrate it to the teaching and learning activities in his classes. Therefore, we figured that it is also important for us to involve not only communities, but also our volunteers who happen to be educators in organizing this program with different guidelines and timelines.
When the pandemic hit the world, all the teaching-learning activities were compelled to be conducted online. Lecturers and teachers were trying and thinking hard to find the perfect way to conduct the online classes. In this case, one of our volunteers came up with an idea to snatch the opportunity to help organize the online learning activities by contributing to Wikimedia projects. We did not think it was impossible, but we were not prepared yet to collaborate with educational institutions to do the online learning activities. Therefore, in next year’s programs, we planned to take a little step by holding a survey, and we will ask the university students to fill it out regarding their needs and the challenges they face during the online learning activities. From there onward, we will try to find the best scheme for WMS to be able to be a part of online learning activities for students, as well as consider its possibilities to be applied in our collaborated institutions.
Last year's Education Internship program (Magang) successfully added 4 new volunteers that contributed not only to the Wikimedia projects, but also to Wikimedia Indonesia’s online activities. Those volunteers were previously interns in our community spaces in Yogyakarta and Padang. In the first semester, we opened 4 slots for the internship program in our community spaces in Padang and Yogyakarta with two slots available for each community. Our community volunteers, assisted by the community coordinator, were in charge of the activities, from the registration, the training, the internship itself, and the final evaluation.
In the second semester, we opened 5 slots for the internship program, 2 slots available for our community in Padang, and 3 slots available for our community in Yogyakarta. Unfortunately, the internship program was just started when the pandemic outbreak happened in Indonesia, therefore we must postpone the program until further notice.
In this year’s program, we managed to complete a guidebook of an introduction to Wikipedia in Indonesian language. This guidebook was aimed to answer people’s questions about Wikipedia and also Wikimedia.
This guidebook was made in two versions. The first version is the less-serious version for the common people, while the second version is the more-serious version for the educators in general. These two guidebooks are now accessible on Commons: General version and Educator version.
Apart from the printed version, we also turned the guidebook into a short video in Indonesian. The actors in the video are Dimas Hardijanto and Fitriayu. Both of them are the contributors of Indonesian Wikipedia and Wikimedia Commons. By creating this video, we hope that we can reach more people to be more aware of Wikipedia and everything that comes with it. From there, we also hope they will join us in contributing to Wikipedia. The video is already available online.
In the next year’s programs, we intend to translate the guidebooks to 3 local languages, they are Javanese, Sundanese, and Minangkabau. We hope that by translating the guidebook, we will be able to reach a broader audience, especially the people who speak those 3 languages, since it will be more relatable to them. We will also include our volunteers to take part in the translation work.
We feel like this year has been overwhelming for us due to the numbers of sub-programs that we tried to complete. This has made us unable to focus on each of the program.
After we held the WikiPelatih PGRI (teacher association) program in Makassar, we kept trying to do some follow-ups, and we also kept trying to assist the teachers in doing their “homeworks”. Despite our assistance and follow-ups, the results of the program did not fulfil our expectations because in the end, the teachers still have not been able to organize Wikipedia editing workshops independently. Thus, their skills of being a Wikipedia editing workshop trainer have not been tested. Their inabilities to organize Wikipedia editing workshops independently was due to the lack of approach and encouragement from us.
We realized that throughout the year, our workshops and training were merely aimed to spread awareness and to introduce Wikipedia to people. However, the program still did not manage to really engage serious participants. Many of the participants disappeared after the events ended. This is probably due to the format of our workshop and training activities which were usually only held once without any follow-up events. Apart from that, our failure to engage more “serious” participants was due to the lack of promotions that we should have been doing through social media as well as during our workshops. We have also notified the participants of WikiLatih and encouraged them to join the Respati program. Some of them did register and try to keep up with the tasks, but many of them failed to successfully complete the challenge.
Expectations and future planEdit
In the next year’s program, we expect to not only be able to introduce Wikipedia to many more people, but also to be able to engage more serious participants to join us in contributing to our online activities as well as our offline activities. These matters become very significant especially when it comes to developing new communities in some regions. Thus, we project to arrange a program that can do the expected works — introducing, engaging, and maintaining the retention afterwards.
To accomplish those expectations, the first step we are going to take is to “shrink” the programs that we will handle a little, so that we will be able to fully focus on our main programs. In next year’s program, we planned to integrate the Respati program with the writing competition, because there are some elements and aspects of the Respati program that actually overlapped with those in the writing competition program. Meanwhile, the internship program will be shifted as one of the community programs, because the outcomes of the internship program have affected the community more on a larger scale than just being a support of the education program. By integrating those two programs into other programs, we will be left with Wikipedia Goes to School, Wikipedia Workshop/Training, and Wikipedia Training of Trainers, the three programs that we will focus on.
Another step that we will take (due to the pandemic situation which hinders us to do offline activities), is organizing online activities until the situation gets better and is safe enough for us to get back on organizing offline activities. These online activities include WikiLatih Online Course, Online WikiLatih, and Online Meetups. Because these activities will be conducted online, we expect these online activities to reach more participants and audiences from different parts of Indonesia whom we have not been able to reach, because we only focused on doing the outreach through offline events.
Once we are in a safer situation and are allowed to hold offline activities, we will try to contact the institutions and or communities who had planned to hold activities with us but got delayed due to the pandemic.
After the pandemic ends, hopefully, we will shift our focus to develop the communities in some regions, by conducting not only once but a few workshop/training activities to help the communities gaining more newcomers. After they have a few active people, just enough to be able to stand on their own and organize online/offline activities by themselves, we will shift our focus to another region.
These steps are in line with our expectations to not only introduce, but to engage, as well as to maintain the retention of the newcomers especially when it comes to developing new communities.
GLAM & PartnershipsEdit
Indonesia has a rich diversity of cultures and knowledge to explore. Today, with the support of technology, culture and knowledge, it should be wide open so that Indonesians can learn their own culture from each other. Limited access to information and knowledge due to lack of technical understanding and other resources should not occur. Since the beginning, GLAM Indonesia has committed to provide full support to institutions that partner with Wikimedia Indonesia to open their collections. The support is in the form of borrowing a scanner, qualified volunteers who have experience in contributing to Wikimedia projects for more than one year, assisting in releasing collections to Wikimedia Commons, and disseminating information through social media to attract public interest to use the collection, and provide opportunities for institutions to be involved in Wikimedia Indonesia activities.
GLAM Indonesia is expected to be able to bring the awareness of the importance of archiving texts, books and other printed media which are vulnerable to damage due to temperature, methods of protection, and media age. GLAM Indonesia helps visionary and aware institutions, which have a shared vision and mission that is in line with Wikimedia to open the access to collections that belong to the public.
For one year, GLAM Indonesia, strived to continue digitization activities that began in 2014 and netted partnerships with new GLAM institutions. This effort resulted in partnerships with several new GLAM institutions and an increasing collection of digitization. As of June 2019, Wikimedia Indonesia continues to partner with eight GLAM institutions:
- Minangkabau Cultural Documentation and Information Center (PDIKM), Padangpanjang,
- Dewantara Kirti Griya Museum, Yogyakarta,
- Yayasan Sastra Lestari, Surakarta,
- Ajip Rosidi Library, Bandung,
- Indonesian Air Force Public Relations and Media Department, Jakarta,
- Sumatran Numismatics Museum, Medan,
- Palm Oil Research Center Library, Medan, and
- Wikimedia Taiwan & National Cheng-chi University, Republic China (Taiwan).
Several other institutions below are in the process of proposing cooperation.
- Riau Malay Customary Institution,
- Riau Province Cultural Agency,
- Pekanbaru City Library and Archives Agency,
- Yogyakarta Cultural Values Preservation Center,
- STAIS Al-Ishlahiyah, Binjai,
- Blanco Museum,
- Puri Lukisan Museum,
- Bali 1928
- Neka Art Museum, and
- Agung Rai Museum of Art (ARMA)
We are proud and grateful for this achievement, thanks to the support of related institutions and the full support of the Wikimedia community.
In addition to collaborating with institutions, GLAM Indonesia took a role in introducing Wikisource, Wikimedia Commons, Wikidata, Creative Commons licenses, and Hack Your Culture event. Several Wikisource workshops have been held in Padang and Denpasar during 2019–2020 and have invited new contributors to proofread texts in Indonesian or local language scripts. Also, we held the 2020 Wikisource Competition, which was participated by dozens of participants to compete for the 2020/2021 WikiNusantara conference scholarship. Wikimedia Commons workshops were conducted which focused on structured data and its close relationship with Wikidata. Wikidata+GLAM workshop was held in Medan City to introduce Wikidata and how to use it within the scope of GLAM. To discuss copyright licensing and collection releases, we held events and activities promoting Creative Commons licenses, as well as in discussions with the GLAM institutions. On January 2020, in collaboration with Creative Commons Indonesia, we held Indonesian Public Domain Day in Medan City using the works of Chairil Anwar and Tan Malaka, which are copyright free since that day. To increase the usefulness of the collections at Wikimedia Commons, we support the Retas Budaya (Hack Your Culture) Festival which involves GLAM institutions and their collections to be adapted into new creative works.
Although we have learned many positive things during the year, we have also learned many lessons that we can use as a basis for strategic planning for this program in the future. We are trying to introduce this program through various channels, including attending several meetings and seminars held by other organizations, with the hope that we can meet with relevant stakeholders to discuss possible future cooperation.
Another thing that concerns us is the lack of awareness of the officials of an institution in Indonesia about the openness of access to collections that should be open. We don't expect it so far to the moon that they can open their collections to Wikimedia Commons, but opening up their database of what collections they have to the public is something we really appreciate. Thus, we can list the collections of each institution we meet to fill Wikidata items. This is made even more difficult by the existence of a manuscript protection law that restricts the public from taking photos without permission from the government, which is granted through certain institutions.
Moreover, due to the coronavirus pandemic, digitalization activities in all offline institutions, which is the core of this program, must be postponed for a certain time. This has had a drastic impact on the progress of digitization, from decreasing the number of scanned collections to postponing workshops which were originally held every month. However, we have considered that some activities, such as a marathon of structured data editing assisted by powerful tools, can be done online. In 2020, we managed to run two structured data marathons to add captions and P180 to the photos of Wiki Cinta Budaya 2020 (Wiki Loves Culture) and Wiki Loves Earth 2020 in Indonesia. This is a new initiative we are taking in response to this pandemic.
In Content Creation, we focus on Indonesian related contents. We improve Wikidata with Indonesian content, building the Javanese OCR, and hold photography and writing competition.
In the last couple of years, the word 'data' has become one of the buzzwords. It tends to be perceived as a 'list of numbers' or things that people often brought up upon in conversation, acting as supportive evidence to a claim. We realized that facing people's perceptions means that we must include the basics of data literacy before teaching people about Wikidata. We used different approaches than Wikipedia training since everyone already has a clear understanding of an online encyclopedia, but not about what is a knowledge base. This year, even though we still focus on giving training and lectures to university students from IT-related majors, we have expanded our volunteers to include people from different backgrounds ranging from GLAM institutions to local communities. We also realized that it is also crucial for us to make our future meetups or training available to be accessed online at a later time because of the pandemic condition, which forced us to postpone all of our scheduled offline activities.
During APG19 (2019-2020), we have managed to hold Wikidata lectures for Duta Wacana Christian University, Andalas University, UPI YPTK University, and Lampung University. The goal of Wikidata lectures is to introduce Wikidata to students with IT background since it is easier for them to grasp the Wikidata concept as most of them are already familiar with the idea of structured data, some of them already understand the semantic web. For the lectures that didn't involve hands-on activities, the student showed an understanding of the Wikidata from the 10 questions quiz after the lectures. Meanwhile, for lectures with hands-on Wikidata editing, the majority of students are able to finish the task of entering data to Wikidata. For our next steps, hopefully, we can organize WIkidata projects or Wikidata clubs with the students.
We have held eight training sessions to different communities based in Bandung, Yogyakarta, Padang, Bali, and Medan. There are two types of the training, the first one is for the basic of Wikidata, aiming for participants who has no or very small understanding about Wikidata, and the second one is advanced Wikidata training, which focused on an in-depth discussion about components and tools related to Wikidata, like QuickStatements, PetScan, and Wikidata Query Service. After the training, we conducted hands-on practices using data related to the community, like universities around the city, famous people born in the city, etc. From this training, we were able to get volunteers with an advanced understanding of Wikidata and also able to help to conduct the training in different cities.
Because of the pandemic, we have postponed several amounts of meetups, lectures, and training. Since our approach focused more on university students, when the pandemic hit, it was hard for us to conduct cooperation with the university. This condition made us realize that we have to prepare online meetups and publish video tutorials online so people can still access and learn Wikidata after the training ends.
We also realized that the lack of trainers made it difficult to reach more areas. We will conduct Wikidata for trainers to create more trainers that can help us spread Wikidata in Indonesia.
Photo competition: Wiki Cinta Budaya (Wiki Loves Culture) and Wiki Cinta Alam (Wiki Loves Earth)Edit
There was a change in Wiki Loves Earth's topic. In 2020, the topic was about protected areas in Indonesia. Photos can be landscapes, flora, fauna, or other objects as long as the object was taken in protected areas, such as in national parks, nature reserves, wildlife reserves, protected forests, and grand forest parks.
This topic is more specific than the previous contest which allowed participants to take photos at any location in Indonesia. Even so, the photos obtained from Wiki Loves Earth 2020 had a better quality than the Wiki Loves Earth 2019.
One of the winners, Fakhrizal Setiawan, said that not all photographers can participate in this contest because the topic is very specific. He got a photo of Piaynemo Seascape Raja Ampat while doing research on coral reef ecosystems in marine protected areas and its relation to food security for dissertation purposes.
Both Wiki Loves Culture and Wiki Loves Earth were conducted online so that we could run these events during the pandemic. There was only a change about the photos' selection by the juries. Prior to the pandemic, they meet in-person to select the best photos. Last time, the juries selected the photos online.
During the Wiki Loves Earth contest, Wikimedia Indonesia received a protest from Jaga Rimba, a youth movement that runs in forest preservation. They have been campaigning and advocating issues on the intervention and exploitation in Indonesian forests, including the conservancy areas. They said that this contest violates Law No. 5 of 1990 regarding Conservation of Living Resources and Their Ecosystems because we allowed participants to take photos in nature reserve areas. They argued if photos taken in nature reserve areas were uploaded freely on social media, it could invite people to visit those areas and contribute to the ecological destruction.
We explained to Jaga Rimba that nature photos on Wikimedia Commons were different from social media photos. On Instagram, people were taking a selfie with certain backgrounds. Meanwhile, photos on Wikimedia Commons were photos of nature itself. Besides, organizing a photo contest cannot be linked to the increasing number of tourists in the conservation areas. Sometimes, photos submitted by participants were stock photos from their past visits.
Wikimedia Indonesia and Jaga Rimba finally agreed to ask the participants about proof of an entry permit to the nature reserve areas if their photos were in the top 10.
Regarding the photo selection in Wiki Loves Culture, we only used the online form. The juries chose photos that were selected by the committee. Then, they gave scores and comments on the photo. This format turned out to be unsuitable because each jury could not discuss determining the best photo. Sometimes, the score given by one jury was very different from the other juries. Therefore, in the Wiki Loves Earth, the photo selection by the juries was conducted through a virtual meeting. So they could discuss and give their opinions directly.
Saraswati (Project GLOW in Indonesia)Edit
Overall there are 1,794 articles approved from Saraswati activities (writing contest, grant, and monthly challenge). However, there are only 1,334 Indonesian articles, the other 460 articles are in the local languages.
Monthly challenge and grantEdit
We were trying to increase Indonesian articles through monthly challenges. This event was announced to the public during April 2020–June 2020. New or existing editors can join. We treated the monthly challenge participants and the grantees differently. Monthly challenge participants are required to make 3 start articles if they want to get internet stipend and 2 C-class articles if they want to get a backpack. They also need to revise the articles, if any, to pass the committee's evaluation. The participants are relatively quick in improving articles. About 1-2 days after the committee sent the notes or revisions, they have sent the article back.
When we look at the grant activity, the condition is different. Many grantees sent revised articles in the following month, and some of them even revised it at the end of the grant period. Sometimes we have to repeatedly remind them about the revisions and encourage them to write new articles. We think it happens because the resources (laptop and internet stipend) are given at the beginning.
Being an active contributor on Wikipedia and other wiki projects does not guarantee that someone has a strong commitment to making grant articles. Only some grantees who make and complete articles based on the requirements. Since April 2020, we allowed grantees to choose the topics. Initially, they were happy and more eager to write. However, after we checked, many articles did not meet the requirements, such as lack of bytes and articles were translations from English Wikipedia. The quality of the articles produced was not as good as the writing contest or monthly challenge articles.
In contrast, the writing participants who are mostly new users can write good articles. At the beginning of the mission, their articles are bad. The committee gives many revisions. Sometimes we have to explain the same things to different participants repeatedly. However, slowly they understand how to write good articles on Wikipedia. They learn from their mistakes. Teaching new users does require more energy. Even so, we are happy to see their progress. After the writing contest ended, some of them were involved in other projects, such as the WikiPerempuan (writing challenge about women), writing challenge on COVID-19 Europe Union, Wikisource, and Wikisosial (write articles for charity).
This year, we have finished developing the software for annotating Javanese characters, which also served an educational purpose to be used by both students and teachers teaching Javanese language at high school. We also made progress integrating our OCR software to be used by Wikisource by making Wikisource's user script so that the user who wants to use the OCR can click on a button in the Wikisource toolbar. Even though the interface for the two machines (our self-hosted OCR software and the Wikisource) to communicate is ready, we still have homework left in terms of processing the inputs from Wikisource and making sure that the software will run at a faster time.
This year's pandemic hit us hard. We were forced to change the coordination between Trawaca's team members and ensure that the team has the infrastructure to do remote work. It also forced us to halt the annotation process because of the restriction to do offline activities. We decided to start building the corpus the transliteration for Javanese characters to Latin characters this year and prepare the annotation process to do it at the next term.
Development of a Javanese learning application programEdit
This application has two purposes, to assist the learning of Javanese characters for high school students and a means of collecting annotation and evaluation data (carried out by local high school teachers and Javanese language experts). After going through the process of using the program to the BOSA Yogyakarta High School in November 2019, we continue to improve the learning application program to accommodate teacher's needs as evaluators. The application has been developed, but we will continue to make some small bug fixes and adjustments in the future.
Development of Javanese OCR softwareEdit
In the development of Javanese OCR software, we refined the program's workflow to make the recognition work smoother than before and accommodate future changes. The program consisted of several modules that do different functions. The steps are as follows:
- Image preprocessing: image binarization, thresholding
- Line segmentation uses the Projection Prole (PP) technique.
- Segmentation of character elements using the Connected Component technique.
- Extraction tour of each character
- An introduction model training that is built with training data generated from the annotation process
- Testing or recognition process by comparing test data (new page input (in real use case).
- Reconstruction of combining character elements into script sequences again
- Mapping character sequences in Unicode
- Display Unicode code on web pages with HTML code as an output introduction.
We have finished making steps 1-6, but we faced difficulties completing the 7th part (reconstruction of Javanese script). This year we decided to change our algorithm to Connected Component, hoping that the segmentation process would improve. But after we implemented it, we faced difficulties related to the amalgamation of the Javanese characters and its components. This was also emphasized by the fact that different scriptures would have different rules of amalgamation as well. Due to this reason, we haven't finished the reconstruction part of this method, and we switched back to the previous algorithm that we used last year (Projection Profile).
Integration of OCR API with Wikisource APIEdit
The construction of the Cakra OCR API was successfully completed in the previous quarter. The test access to the API service was successfully carried out in a cross-domain manner with a test call from the partner domain https://ferianto.id/twclient/. This API service manager module is based on the Yii framework and can generate a unique token for each domain that making the call. During this quarter, tests for API calls were carried out periodically. There was a problem where the test code at https://ferianto.id/twclient/ could not make an API call (unauthorized) even though it had the appropriate token code. After investigating, it turned out that the token validity duration was very short, and it was temporarily disabled until a better management method was found.
This year, the integration of the Cakra OCR API service has been successfully carried out with the help of volunteers and WMF staff at Phabricator, namely Aklapper, Xover, and Tpt. We are very thankful for their help. The integration steps were successfully carried out, namely (1) creating OCR.js code as API caller and receiving code, then (2) generating common.js code for initializing OCR.js calls, and (3) test interface functionality.
The test for calling and receiving Cakra's API OCR answers went well, but we just realized the API return report code states that the data sent from Wikisource is a BLOB. For this reason, we are going to investigate the suitability between the API design that initially received the call input in the form of a JPG image, and whether this is one of the causes of the empty OCR result.
Construction of a syllable corpusEdit
To support the transliteration program that will be implemented next year, we have created a syllable corpus which is done by exporting the reading column values from the database of the Javanese script learning application program and the script annotation program from the APG18 project. This export process certainly does not experience significant difficulties. The process of building a syllable corpus is extended by writing a program that breaks words into syllables. The program code is built by applying the Finite State Transducer (FST) model.
First is the effect of the COVID-19 lockdown to the development of the Javanese Script Learning Application. The lockdown is very influential on the development rate because our team is accustomed to physically working on application code together to make the synchronization easier. Remote working turned out to be an unpredictable obstacle because some of us were struggling to get a good internet connection. Even though we have done online meetings every Tuesday and made our self-hosted sandbox to do tests and synchronize our code, we still feel that it is not enough for us to coordinate properly. Likewise, the absence of face to face increases distraction so that in coordination meetings, sometimes we have to check the same things over and over again. The lockdown also forces us to postpone the testing and annotation process, which we originally planned to do with Immanuel Kalasan High School and Budya Wacana High School.
The second lesson learned was selecting a new segmentation method (Connected Component Labeling) had unexpected consequences. What should have been expected to improve the OCR's quality (leaner class, more consistent data) turned out to have added a more complicated effect on recombining the Javanese characters. We learned that something that works would be more useful than something that is ideal but is hard to use, which goes back to our original vision with Wikimedia Indonesia when we started this Javanese OCR Project. Thus, all milestones for the following year will be based on this vision alone. We decided to switch back into our previous detection method, namely, Projection Profile techniques, for line and script segmentation. The performance improvement of the OCR system's recognition will be carried out at the preprocessing stage as an alternative to doing edge detection. We also decided to develop a semi-supervised learning model to increase the number of character annotations used as training data, impacting the accuracy of the recognition model.
The next thing that we learned is that we have to adjust the Trawaca API, because even though it has been sufficient for internal use within TRAWACA.ID, but it is still not able to be used properly in Wikisource. This is due to (1) sending document data from Wikisource in the form of BLOB not JPG images, and (2) the return API must be in plain text, while Cakra's JSON API structure currently places the OCR results in the
<nowiki> </ span tag ></nowiki> and we have to change it into plain Unicode for Javanese characters. Thus, in the following year, we will focus on improving the API according to the above specifications.
The final lesson learned was shifting the approaches that previously work very well into something more advanced may not always lead to good results. In the scope of our projects, this has happened in two cases. The first one was for the Connected Components, and the second one was when we try to use deep learning for object detection, and it turned out that deep learning requires a graphics card process with CUDA Core, which the team currently doesn't have. Apart from unfulfilled needs and a longer process, we finally realized that the initial approach taken with a statistical approach was proven to be sufficient to detect non-text objects (images or illustrations), with the average shape found in Javanese script
Revenues received during this 12-months periodEdit
- S1 is per 31 December 2019
- S2 is per 30 June 2020
|Revenue Source||Currency||Anticipated||S1||S2||Cummulative||Anticipated (US$)||Cummulative (US$)||Explanation of variances from plan|
|FDC APG Grant||IDR||4,292,505,000||2,494,945,810||1,797,559,190||4,292,505,000||304,767.86||304,767.86|
Spending during this six-month periodEdit
Please note that the USD rate is the proposal rate. During the Pandemic, especially from March to April 2020, the IDR rate plummeted drastically from USD, so for the actual spending, you may refer the IDR amount.
|Expense||Currency||Budgeted||S1||S2||Cummulative||Budgeted (US$)||Cummulative (US$)||Percentage spent to date||Explanation of variances from plan|
|Office and administrative support||IDR||705,000,000||375,567,489||447,410,274||822,977,763||50,055||58,431||116.73%||We moved our offices, redesigned the office, and bought various items to follow the government regulations about COVID-19|
|Salaries||IDR||1,185,325,000||554,041,304||654,841,667||1,208,882,971||84,158||85,831||101.99%||Additional manpower for short term projects (Respati) and additional tax|
|Community Building and Support||IDR||796,600,000||166,673,588||235,685,454||402,359,042||56,559||28,567||50.51%||Pandemic effect caused cancellation of various events|
|Education||IDR||752,080,000||468,620,172||182,558,859||651,179,031||53,398||46,234||86.58%||Pandemic effect caused cancellation of various events|
|GLAM||IDR||366,500,000||77,319,420||104,469,840||181,789,260||26,022||12,907||49.60%||Pandemic effect caused cancellation of various events|
|Content Enrichment||IDR||285,000,000||48,080,741||126,509,601||174,590,342||20,235||12,396||61.26%||Pandemic effect caused cancellation of various events|
|Program Support||IDR||202,000,000||63,332,855||508,270,116||571,602,971||14,342||40,584||282.97%||We changed the plan during the pandemic and shifted the expense to increasing the staff's capacity building and wellbeing. We enrolled them to join the online English class, additional benefit of private health insurance, and various online class.|
Is your organization compliant with the terms outlined in the grant agreement?Edit
As required in the grant agreement, please report any deviations from your grant proposal here. Note that, among other things, any changes must be consistent with our WMF mission, must be for charitable purposes as defined in the grant agreement, and must otherwise comply with the grant agreement.
Are you in compliance with all applicable laws and regulations as outlined in the grant agreement? Please answer "Yes" or "No".
Are you in compliance with provisions of the United States Internal Revenue Code (“Code”), and with relevant tax laws and regulations restricting the use of the Grant funds as outlined in the grant agreement? Please answer "Yes" or "No".
- Once complete, please sign below with the usual four tildes.