Data Visualization Community Group Meetup Event Minutes

20 Sep 2015

See also: IRC log


Chunming(W3C), Angel(W3C), Xiaoru Yuan(Peking University), Xin Huang(Qihoo360), Renhua Liang(Zhejiang University), Dahua Guo(Chinese Academy of Sciences), Jie Liang(Peking University), Limei Che(Baidu), Yueli Jun(Haiyuan Data), Yanbo Chen(Peking University), Yi Chen(Beijing Technology and Business University), Tiemeng Li(Beijing University of Posts and Telecommunications), Jinghua Zhao(China Electronics Standardization Institute), Renjie Cao(Zhongqizhiyuan, publishing_house), Jiajun Tang(GAPP), Deqing Li(Baidu), Hanchen Song(National University of Defense Technology), Yadong Wu(Southwest University of Science and Technology), Yuan Ren(Caixin), Shenyuan Wei(Caixin), Ming Lu(Peking University), Shan Chen(Huawei), Xiang Li(Wuhan University), Kai Rao(China Mobile), Lijing Li(Peking University), Qian Huang (SuperMap Software)
Xiaoru Yuan (Peking University)


Welcome from the Event Host and Round Table Introduction - Xiaoru Yuan (Peking University)

Xiaoru Yuan: I am Xiaoru Yuan from Peking University. Welcome to Peking University! We are glad to have you all here today. Now let's get warm up by the round table introduction.

[round table introduction]

W3C Standards Making Process and Its DataVis CG - Chunming Hu (W3C)

Chunming Hu: Hi all, as introduced, I am Chunming Hu from W3C/Beihang
... today I will introduce the way W3C standards are made and its DataVis community group
... please find my slides at http://www.w3.org/2015/Talks/datavis-meetup-presentation-20150920

[Chuming go through his slides]

Thoughts on Datavis Standards Work - Xiaoru Yuan (Peking University)

Xiaoru Yuan: Thanks Chunming for the detailed intro! Now I would like to share my thoughts on why we should do the data visualization standards work. There is only one page on my slides. I started the data visualization research work 15 years ago. Data visualization as a research subject has been developed for about 26 years. in the year of 1988 there was a small meetup in the US discussing the data visualization work, just like the event we are having today. The topic of data visualization was proposed and discussed by a group of researchers, which can be considered as the starting point of data visualization research. The data they were discussing at that time was mainly about data in the science and technology world. Nowadays, the data in terms of data visualization has been introduced into a much more broader community and various industries and fields. Waves of new technologies like HTML and XML support data visualization.

Xiaoru Yuan:The research work on data visualization in China has been more and more promising in the recent years, which can be seen from the increasing amount of related papers published in the data visualization conferences. In the IEEE Visweek conference last year, there were 89 Chinese participants. 5%-10% papers accepted by the event were from Chinese. Also, big Chinese IT companies like Qihoo360, Tencent, Baidu and Alibaba have been involved in this field as strong industry support. With these promising aspects as the background, I started the conversation with Chunming from W3C/Beihang about datavis later last year. After over 9 months preparation, now I think it is time for us to start the work. The standardization work of data visualization needs strong involvements from Chinese research community and industry. To give it a harder push, I've proposed to China Computer Federation that a data visualization committee should be set up under its umbrella. 

Xiaoru Yuan: We can see two clear trends in the datavis world. One is the datavis work spreading from people with expertise to the public, while the other one is datavis moving from native to the web. And that is why I think relating datavis standards with Web should be a proper starting point. As for the standards work for datavis we are about to discuss today, it does not have to be limited to the web. I think we should hear a broader voice from the academic circle and the industry, e.g. the requirements for S&T data visualization can be taken into consideration as well.

Xiaoru Yuan: For the standard work of data visualization, terminology can be a good start. There have been some quite some good terminology about data visualization in English, but not many in Chinese. A unified terminology is very important for us to start. For the same data, there can be various ways to visualize it. However, if there is no standard APIs, it is very difficult to integrate them.

Xiaoru Yuan:  Besides the enthusiasm we see from today's event, there are international peers who would like to join the standards work for data visualization too. So today, I hope we could get ready to start the work and be prepared for a long term journey. Thank you all for participation! Your contribution to datavis standard work is really appreciated.

Frond End Development Requirements for Data Visualization Standards - Xin Huang (Qihoo360)

Xin Huang: Hi all, my name is Xin Huang from the data visualization team of Enterprise Security Department of Qihoo360. Today, I would like to share some problems or so called requirements from our practices. My team has been working on data visualization for 3 years and we have been cooperation with Peking University on this topic since 2012. First, let me show you some demos of the datavis work of my company.

[Xin Huang go through some demos on security attack, DDOS, collaborative interactive analysis tool, etc]

Xin Huang: Here I will share some requirements for data visualization from our work. First, SVG for simpler and easier responsive interaction. It is important to provide better service to the mobile devices. Now we have to take both viewBox and preserveAspectRatio into consideration. Second, richer semantics for presentation, e.g. better presentation for the associated diagrams. Third, standard APIs. Web engineering is becoming larger and larger with HTML, CSS, JS, etc. CSS should not be the only one to style pie chart or bar chart. Web components should be independent. Fourth, more powerful web app capabilities, e.g. GPU cloud rendering on the web for unified user experience. 

Echarts Next - Deiqing Li (Baidu)

Deqing Li:Hi all, I'm Deqign Li from Baidu. Today the topic I would like to share with you is Echarts Next. Comparing to 3D, Echarts has certain advantages, e.g, easier to learn, with lower cost. We've been receiving feedback from users wrt various problems. Our users have various requirements for datavis via Echarts. When we use the traditional way to visualize data, the style is too complicated to present different charts. The use cases we have so far are mainly about 1D and 2D data, not so much for 3D which is more different to present. There are some 3d party extensions. With the templates of these extensions, we can have some process, e.g. data(model, format) -> category (dimensions, layers, time sequence, geospacial, etc) - > filtering -> mapping from data to diagram (glyph,size,color) - > coordinate system/layout - > interaction (webGL, canvas, SVG).

Introduction of National Standards Work on Big Data - Jinghua Zhao (CESI)

Xiaoru Yuan:The State Council has announced the strategy for big data research, which is a very important topic for next step of national standard system. So we are very glad to have Ms.Jinghua Zhao from CESI to share with us about the national standard work on big data.

Jinghua Zhao:Hi, I am Jinghua Zhao from CESI.  Glad to be here to share the big data standard work lead by CESI. CESI has been doing some research work on the standardization of big data world wide. There are many international standard bodies engage in this topic. Generally speaking, some main standard bodies working on big data include ISO/IEC JTC1 WG9, CS32, ITU and NIST.  The work in ISO/IEC JTC1 WG9 is about the basic standards for big data which includes ISO/IEC 20564, 20567-1, etc. WG9 will meet in Beijing in July next year. ISO IEC JTC1 SC32 is working on data management and exchange, which include metadata, model and ontology. ITU-T13's has some work related to the data on WoT. NIST NBD-PWG has 5 WGs and 7 outputs. As for the standardization of big data in China, it is mainly carried out in the BDWG, with 7 sub-groups and 148 organizations participating. A whitepaper V2.0 of Big Data Standards will be published by BDWG soon. your comments are welcome! BDWG is also in a joint effort with ISO/IEC JTC1 WG9.

[Jinghua shows the organizational structure of DBWG]

JInghua Zhao:There are 10 big data standards projects launched under BDWG. The part of Terminology and Reference Model has been released to call for comments. These projects also include data capability maturity, semantic description, reference for scientific data, data trading platform, etc. Data visualization is an important part of scope too. Requirements on datavis has been discussed within the BDWG, though not much progress has been made yet. Datavis standard work is included in the application layer of the big data platform framework 2.0 as well. So the communication and close cooperation with data visualization community matter to the standard work of BDWG. We would like to work with you all on the requirements, standards making, testing and topics related. Looking forward to future cooperation! 

Open Discussion - All

Xiaoru Yuan: We have proposed some topics for this session, e.g. requirements, use case, scenario, roadmap for data visualization standards work, possible work scale and work plan , etc.

Yuan Ren (Caixin): As a content provider, the datavis work of my company kind of differs from the work of Baidu or Qihoo. They are more tools oriented while we are more focusing on the content and data. Our user use data from Excel and CSV files. How do we present them via tree structure or network structure? Our employees are mainly editors and reporters with limited computer skills, so we would like to have some standard ways to make data visualization easier for them. One example, we did a long report about Yongkang Zhou. To explain the complicated relationship within his group, we used Echarts and some other tools. However, they could not meet our requirements. It would be really interesting to see what areas of data visualization will be standardized. As for us, we would want something more down to the bottom, e.g. the algorithm.

Yadong Wu (Southwest University of Science and Technology):We should consider more aspects of the requirements for datavis standards, besides those from the industry, the requirements from government or public affairs should be considered too. Huge amount of gov info data needs to be visualized, e.g. the transportation, tourism, etc.

Limei Che (Baidu):I had 3 years of study on datavis, got some ideas on datavis model and algorithm. Now I am working in Baidu. The practice we have in our products match what Prof.Wu just said. The use case we have is our product called Baidu Tourism, which involves semantic analysis, interaction analysis, etc. From this use case, we are think whether we should standardize more elements, e.g. parallel coordinates and primitives. Data visualization as a technique is getting closer and closer to the public, and there should be more education materials to promote it. Another suggestion is about standards for UE. Better cooperation and communication between datavis designer and UE designer is needed.

Qian Huang (SuperMap Software): I was involved in some national standard and I would like to share my thoughts on datavis standards. A good standard should always evolve and we should not expect to finish the standards work on datavis with one efforts. As for the terminology, it makes sense to even standardize the title of the positions involved. Data visualization is crossed with many fields, e.g. art and S&T. So targets/goals of datavis should be very clear and feasible. A proper working scale is very important. The work on Dynamic visualization is almost empty, which might need our attention.

Dahua Guo (Chinese Academy of Sciences): My team works on a national project on data integration. 2 challenges faced, one is how to integrate the data, other is to visualized them. If there are standards for them, it will make our work much easier. For data visualization standards, the basic tool level can be standardized, but not the application level. 

Yi Chen(Beijing Technology and Business University): Some thoughts for the meeting today. 1st, we need to let the industry know the importance of the datavis standards, why we are doing it; people did not think it datavis could solve problems before big data got hot. 2nd, some common ways like the pie chart are already well known but more new ways are coming, which is not well known to the public yet and some best practices to promote these new solutions would be needed. 3rd, my team works on the datavis of food safety, which is more certain industry focused, I need to see how my work match this datavis standards.

Jiajun Tang (GAPP): When we are making standards, besides national and industrial standards, consortium standards and company standards are important too.  When an organization is working on its own standards, there are some really good requirements and practices we can make use of. For datavis standards, the requirements from companies matter as well as their proposed solutions. That is to say, the standards making should be bidirectional. One task I bring to this meeting is that GAPP is working on the 13th 5-year plan wrt to the press and publishing industry. And data visualization is part of the scope we need to look into. One challenge for us is to indentify what is the proper description of big data for press and publishing industry. And I would appreciate if we could have some opportunity to have a further conversation on this some other time. 

Ronghua Liang (Zhejiang University of Technology): we might need some industrial standards for data visualization as well, and we need to avoid the overlap work with the work of DBWG.

Wrap Up

Chunming Hu:  I hear it that we have got the consensus it is a good time to make the standards for datavis and some good requirements have been collected from the presentations and open dicussion. The possible working scope of datavis standards might include terminology, data model, core standards on meta data, visualization events and interaction, style of visualization method, and some common visualization methods based on the metadata and so on. As for the related national standards work, the BDWG and 13th 5-year plan of GAPP might fit, which I hope would lead to futher conversations with related parties. I would like to invite you all to join W3C datavis community group which can be a good platform for us to continue the conversation.

Xiaoru yuan: Thank you all for the active participation in the discussion!  Good time control. For the future work, we will discuss with the community and choose carely what scale we should take. It would be very exciting and helpful to make some basic standards to support the datavis work. Thank you all again for coming to this event today! More conversations will be carried in mailing list. Welcome to join the Datavis CG and look forward to working with you all!

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.140 (CVS log)
$Date: 2017/04/25 06:40:22 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.140  of Date: 2014-11-06 18:16:30  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Succeeded: s/been announced/announced/
Succeeded: s/NIST has 5 WGs and 7 outputs/...NIST has 5 WGs and 7 outputs/
Succeeded: s/Baidy/Baidu/
Succeeded: s/Angel_/Angel/
Succeeded: s/attendtion/attention/
FAILED: s/attendtion/attention/
Succeeded: s/field/fields/
Found Scribe: Angel
Inferring ScribeNick: Angel
Present: Chunming Angel Xiaoru_Yuan Xin_Huang(Qihoo360) Renhua_Liang(Zhejiang_University) Dahua_guo(CAS) Jie_Liang(Pekeing_Uni) Limei_Che(Baidu) Yueli_Jun_(Haiyuan_Data) Yanbo_Chen(Peking_Uni) Yi_Chen(Beijing_Gongshang_Uni) Tiemeng_Li(_Beijing_Post_Uni) Jinghua_Zhao(CESI) @@@(zhongqizhiyuan publishing_house) Tang_(GAPP) Deqing_Li(Baidu) Hanchen_Song(Guofangkeda) Yadong_Wu(Xinankeda) Yuan_Ren(Caixin_Data) Shenyuan_Wei(Caixin_Data) Ming_Lu(Peking_Uni) Shan_Chen(_Huawei) Xiang_Li_(Wuhan_Uni) Kai_Rao_(China_Mobile) Lijing_Li(_Peking_Uni)
Got date from IRC log name: 20 Sep 2015
Guessing minutes URL: http://www.w3.org/2015/09/20-datavis-minutes.html
People with action items: 

[End of scribe.perl diagnostic output]