Crowdsourcing/Crowdsensing Leveraging the Power of the Crowd in Science

National Institute of Informatics News ISSN 1883-1974 (Print) ISSN 1884-0787 (Online) NII Interview Crowdsourcing Will Change Society 56 Exploring Structure Hidden in Dec. 2015 Real-World Data Using a Game in Quantum Computer Research Crowdsensing Useful in Solving Social Problems Feature Crowdsourcing/Crowdsensing Leveraging the Power of the Crowd in Science

This English language edition NII Today corresponds to No. 70 of the Japanese edition NII Interview Crowdsourcing Will Change Society What are its merits, and what challenges will accompany its popularization?

Dr. Vili Lehdonvirta (Research Fellow and DPhil Programme Director, Oxford Internet Institute, University of Oxford） Interviewer: Keiichi Murayama (Senior Staﬀ Writer/Editorial Writer, Nikkei Inc.)

The worldwide Internet population is ap- Murayama: With crowdsourcing service competition has intensified due to interna- proximately 3.2 billion. The environment providers listing shares in Japan, crowd- tionalization, trade liberalization, and so connecting this vast number of people is sourcing is spreading on a global scale. on. Companies are exploring new forms of changing our economy and society, in- Lehdonvirta: I began researching digital la- employment in order to remain competi- cluding how people are employed and bor, or crowdsourcing, in 2010. There are tive and enhance shareholder value. how they work. Crowdsourcing has be- various types of crowdsourcing, but letʼs Murayama: There are also technological come a global trend that connects busi- talk about crowdsourcing for business factors behind the spread of crowdsourc- nesses who want to outsource work and here. This ranges from outsourcing of rela- ing, perhaps? people who want to take on work via the tively simple work to use in R&D. Lehdonvirta: The speed and quality of the Internet. What influence will crowd- Looking at a time span of several de- Internet has increased. In the past, I have sourcing have on business organizations cades, the whole concept of employment interviewed 127 people who earn their liv- and innovation? I asked Dr. Vili Lehdon- in economically developed countries such ing working online as freelancers. These virta, a Research Fellow at the Oxford In- as Japan, which until now was stable and were workers in developing countries such ternet Institute in the UK. permanent, is crumbling. Employment pe- as the Philippines, Malaysia, Vietnam, riods have shortened, and flexible, contrac- South Africa, Kenya, and Nigeria. The Inter- tual employment has increased. Global net and mobile infrastructure are spread-

Research Fellow and DPhil Programme Director, Oxford Internet Institute, University of Oxford. PhD (Economic Sociology). Conducting Vili Lehdonvirta research on the social and economic signiﬁcance of new information technologies, focusing on virtual economies and online work. Visiting Scholar at the Interfaculty Initiative in Information Studies, University of Tokyo, from 2010 to 2011; Visiting Graduate Student at the Faculty of Science and Engineering, Waseda University, from 2006 to 2008.

02 Feature │ Crowdsourcing Will Change Society ing in these countries too, and crowdsourc- smaller as a result. scribed as fair because it is capable of re- ing is becoming possible. However, there are various reasons for ducing discrimination by applying rules to However, there are problems associated the existence of business organizations. In ensure that employees are promoted with crowdsourcing because it easily cross- particular, there are some effects that can- when they acquire the necessary abilities es over national borders. If employment is not occur without the business organiza- and skills, regardless of gender or race. concluded inside a country, the respective tion, for example experiencing an identity Whereas, if this is left to the free market, as government can protect workers by way of and sharing a vision, which can increase in crowdsourcing, there is a risk that dis- the legal system, but this is not the case for motivation. Therefore, I do not think that crimination will occur. In fact, looking at crowdsourcing. It is not easy for a country we will reach a point where all employees the US car dispatch app Uber Technologies, to demand that another country apply the work as freelancers. it appears that black drivers are more likely same labor contracts as it does, and in any Murayama: R&D is a more important opera- to get low ratings. In a taxi company, such case, workers may be anonymous and im- tion as far as business is concerned. Do you discrimination could be controlled by way possible to identify. There is also the poten- think the time will come when crowd- of company rules. The risks of relying on tial for opportunism, such as fraud and sourcing is commonplace in R&D? market forces should not be overlooked. cheating, to become prevalent. Lehdonvirta: The main benefit of using Much more research is needed. Murayama: What should be done? crowdsourcing in R&D is the effect of open (Photography: Yuki Akiyama) Lehdonvirta: At this point, one potential innovation. Conventionally, a companyʼs solution is something called a “reputation ability has been restricted to the abilities of system”. The US Internet auction site eBay the staff employed in the company, but us- was the first to fully introduce this system, ing the Internet, it becomes possible, in and the idea behind it is to maintain quali- principle, to access all abilities. Various ty by having participants evaluate each ways of thinking can be obtained from a other. diverse group of people, compared to a ho- However, although it appears simple mogeneous group. and straightforward at first glance, there One point that requires attention, how- are difficulties. For example, people some- ever, is whether companies are capable of times award, despite their real opinion, the absorbing and making effective use of A Word from the Interviewer highest rating of five stars even though technology and ideas taken in from out- they were not satisfied with the service side. Ideas for boosting productivity by they received because they fear that the 10%, for example, may be realized by mod- other person may retaliate if they give ifying existing equipment or procedures, them a low rating. This is because it is a but more fundamental ideas could require “tit-for-tat”[Literal translation is “give-and- changing the entire business model. take”] system. The problem is how to de- This is not simply a matter of it being rive genuine, rather than collusive, ratings. okay to change everything using open in- Also, it is not easy to start accumulating novation. In addition to winning future ratings from zero and compete with peo- customers through innovation that re- With the ability to serve the diverse needs of individuals in relation to ways of working and ple who already have many reviews and forms their business model, companies the potential to become a driving force for have earned a high rating. The structure is have to serve their existing customers. It is corporate innovation, there is no doubt that crowdsourcing holds great promise. Howev- such that disparities arise between individ- important to strike a balance between cur- er, as Dr. Lehdonvirta: repeatedly emphasizes, uals. There is room for improvement in the rent and future customers. it also conceals new problems and challeng- whole system of crowdsourcing, and we Murayama: In Japan too, values have diver- es. While closely monitoring the situation, we must acquire the knowledge to master need to come up with good rules. sified and various ways of working are now crowdsourcing. Murayama: How will crowdsourcing accepted. Do you think that the addition of change the shape of business? crowdsourcing as an option should be Keiichi Murayama Lehdonvirta: The spread of the Internet has viewed positively? Joined Nikkei Inc. in 1992. Covered telecom- radically changed the structure of employ- Lehdonvirta: It would seem preferable to munications, electronics, automotive, fi- nance, etc., in the Business News Depart- ment costs. It has become easier to imple- have more options, but of course, there are ment. After working in the Silicon Valley ment network-based employment as an complex problems involved. When a com- Bureau and the Electronics News Depart- alternative to the conventional hierarchical pany is referred to as bureaucratic, it cre- ment, became a senior staff writer in April 2012. Has worked concurrently as an editori- system of employment, and it has been ates a negative impression, but in a sense, al writer since April 2015. Currently, his main suggested that companies will become this kind of company could also be de- areas of responsibility are IT and startups.

2015 NII Today │ No.56 03 Exploring Structure Hidden in Real-World Data Aiming to promote value creation and citizen science

Asanobu Kitamoto (Associate Professor, Digital Content and Media Sciences Research Division, National Institute of Informatics/ Associate Professor, School of Multidisciplinary Sciences, The Graduate University for Advanced Studies)

Weather disaster prevention, participa- an interactive digital archive that enables example, precipitation amounts can be tory monitoring of invasive species, cul- users to create and display virtual exhibi- found using existing sensors. However, ex- tural heritage archives, memories of the tions using images of Silk Road cultural isting sensors cannot give an answer to Great East Japan Earthquake... making heritage to be printed as an original pic- whether precipitation is rain or snow. This use of not only conventional observation ture postcard. information is currently estimated from air data collected by specialists but also data The common theme in all of these proj- temperature at the ground surface. How- collected by the general public will reveal ects is “generating value using real-world ever, if it were possible to gather com- the structure of a previously unseen data”. Real-world data are data that link to ments, such as “Itʼs started snowing” or “Itʼs world. NII Today spoke to Associate Pro- reality, including unstructured data from sleeting”, from the public via social media, fessor Asanobu Kitamoto, who is trying the public. Associate Professor Kitamoto is and to map these comments in chronolog- to open up a new world by tackling di- attempting to generate valuable informa- ical order, we would obtain information verse subjects using image processing, tion and present it as media that takes ad- that has never existed before. technologies for utilizing big data, and vantage of new technology. For example, Tools for public participation are being citizen science. in the GeoNLP Project, he is conducting re- developed. “Memory Hunting” is a smart- search on automatic extraction of place phone camera app that enables users to Raising awareness and gaining names from text and linking place names take a photograph at the same location learning opportunities to locations. Allocating[Consider “Extract- and same angle as a previously taken pho- ing”.] location information is an important tograph. It was originally developed as part Associate Professor Kitamoto has step in making people take action, particu- of the Digital Silk Road Project for finding worked on a wide range of different proj- larly during disasters. Rather than simply locations where old photographs had been ects. These have included “Digital Ty- showing information to people, Associate taken. The use of an app that allows multi- phoon”, an integrated database of ty- Professor Kitamoto aims to create media ple photographs to be taken with the same phoons including images from weather that convey messages or stories that trans- composition makes it easy to carry out satellites to measure typhoons, typhoon late into action. fixed-point observations. It enables infor- tracking data, and textual information One might suspect the value of data ob- mation to be acquired more widely and from social media about typhoons; “Ty- tained from the general public. These data from any area, in addition to fixed-point phoon Front”, a user-generated typhoon are, of course, pointless if they are just a observations from satellites and ground- information site; and “Senga—Silk Road”, poor manʼs version of existing sensors. For based observations from existing sensors. However, Associate Professor Kitamoto explains that citizen science is not simply a device for collecting data. In crowdsourcing, the mission is to complete a task, but in citizen science, the citizens who participate are not workers but collaborators. For them, citizen science is a learning opportu- nity that raises their awareness of their sur- roundings and brings the world into sharp- er focus. Also, there is value in it if citizens become able to consider the whole picture. Sharing of goals is important Figure1 “Seiyou Status”, a citizen-based monitoring project to control an invasive species. Ongoing activities are being carried out in cooperation with the Laboratory of Conservation Ecology at the Univer- sity of Tokyo (currently, the Hokkaido Biodiversity Conservation Activities Cooperated Support Center) to control The “Seiyou Status” project to control an Bombus terrestris, which is designated as an invasive alien species. Reports of sighting and capture are published on a website in list and map formats. invasive species in Hokkaido is a typical cit-

04 Feature │ Crowdsourcing Will Change Society izen science project. Focusing on citi- sors, but information about how residents zen-based monitoring, the project publi- feel about the subjective risk of the water cizes information on the capture and level of the river can only be obtained via sighting of an invasive species of bumble- people. bee, Bombus terrestris. The project runs Proposing new methods of short courses locally to teach people how research to capture the bees and what to do if they Figure 2 “Memory Hunting”, a smartphone app get stung, and gathers capture results to Research handling real-world data has that enables users to take a photograph at the same location and same angle as a previously show the invasion status on a map. Ecolo- its own particular difficulties. For example, taken photograph. gy researchers can use this information to the fields of meteorology and informatics examine how the bee spreads and how have different ways of approaching re- crossed, truly interesting discoveries will land should be used, and citizens become search questions and solutions. Associate never be made. more aware of the problem of invasive Professor Kitamoto deals with past data Associate Professor Kitamoto says that species. using informatics techniques and carries what informatics can do in such circum- The important thing is not just whether out research from the perspective of how stances is “propose new methods of re- there are people who can get citizens in- to use these data in todayʼs decision mak- search.” volved but also whether the goal of the ing. Meteorologists, on the other hand, Open science and citizen science are ini- project is shared. “People will not partici- rarely do this kind of research because they tiatives for exploring new possibilities to- pate unless there is a shareable goal.” In are interested in the actual weather phe- gether, while promoting mutual under- the case of “Seiyou Status”, the goal is to nomena themselves. standing. protect the ecosystem. However, collaboration is necessary. There (Interview/Report: Kazumichi Moriyama, There is also a project planned in Hokkai- are limits to the meaning that informatics Photography: Kazuyoshi Oyama) do this winter to observe snow conditions researchers alone can draw from data. based on citizen participation making use “We lose the potential of research if we of the “Memory Hunting” system for taking regard it as research confined to informat- photographs with the same composition. ics, but if we do it together with meteorol- However, when the goal is disaster pre- ogists, we must find common themes.” vention, despite being a goal that is easily Carrying out research across disciplines shared, it is difficult to sustain motivation is not easy, because both sides have to among participating citizens. It is fun to take a completely new approach to their observe phenomena that cause big chang- research. However, unless disciplines are es in a short time, such as observing the flowering of the cherry blossoms as a seasonal change, but when it comes to disaster prevention, most of the time the situation is normal and nothing happens. This makes it easy to forget the goal of disaster preparation. In this context, Associate Profes- sor Kitamoto is focusing on “emotion sensing” as a way to observe citizens using their comments. Emotions are not objective: they can only be obtained through people. For example, information such as the water level of a river can be obtained using existing sen-

Asanobu Kitamoto

2015 NII Today │ No.56 05 Quantum Computer Game Computer game can assist the research to realize quantum computers

Kae Nemoto (Professor, Principles of Informatics Research Division, and Director of the Global Research Centre for Quantum Information Science, National Institute of Informatics/ Professor, School of Multidisciplinary Sciences, The Graduate University for Advanced Studies)

Professor Kae Nemoto and her research There are two aspects to the game de- vancing research by helping compile and team have developed a computer game veloped by Professor Kae Nemoto and her optimise quantum algorithms,” says Pro- called “meQuanics” that allows the gen- team, meQuanics: the quantum computer fessor Nemoto. eral public to help optimise real-world game. One is the actual game itself, which “Reducing the size of circuits makes it quantum computational circuits and re- is directly related to how a quantum com- possible to reduce the size of quantum duce the number of devices in a future puter will be built and programmed. The computers, shorten processing times and quantum computer. They have released other is to allow the game to use ultimately making the machine easier to a prototype client that can be played and “crowd-sourcing” to help make circuits build. By making the circuits in meQuanics tested online. When a player starts the smaller, faster and easier to implement. smaller, you can have the same effect as game, a puzzle consisting of colorful, Crowd-sourcing is the process of using the asking experimentalists and engineers to tangled shapes appear. By twisting and resources of individual people to help solve make devices ten times more accurate. shaping the puzzle, players make them complicated problems. meQuanics is based Since the problem in meQuanics doesnʼt smaller and optimise resources for an ac- around a model of quantum computing［1］ require any fancy equipment or expensive tual quantum computer. The team aims known as topological computation, which manufacturing, this could be of enormous to attract a large user base of active play- is currently the most promising architectur- benefit.” ers using smartphones and tablets and al framework for building quantum com- The interesting thing about meQuanics uses the power of crowd-sourcing to ad- puters, a reality and a primary research top- is how accurately we can reflect the physics vance their research into large-scale ic in the research center led by Professor and engineering within a gaming environ- quantum computers. Nemoto. ment. “The puzzles actually represent the Promoting research with a “The primary aim with meQuanics is to circuits themselves, we donʼt have to ap- puzzle game build a game that lots of people can enjoy proximate the physics or the information playing and by playing, you are actively ad- theory itself. Hence the compilation and optimisation itself links naturally to solving a 3D puzzle game. We were lucky in that the problem itself contained puzzle-like el- ements,” reflects Professor Nemoto. A characteristic of the topological quantum computers that Professor Nemoto is researching is that crucial error correction protocols, needed to compensate for inac- curacies in the quantum hardware, is built into the formalism. We encode one quan-

Figure Screenshot of the tum bit (qubit) of information using hun- meQuanics game dreds, maybe thousands, of physical qu- Players reduce the size of the entire puz- bits. It is this redundancy that protects the zle by moving, cutting, and connecting puzzle pieces in 3D. information in the computer. Algorithms References are executed by manipulating a large 3D array of qubits in a topological way, and ［1］quantum computing Quantum computers have the potential to dramatically outperform the “classical computers” in a wide variety of import- this representation of an error corrected ant problems. Experimental construction and demonstrations of quantum computers is happening throughout the algorithm can be represented as a 3D puz- world, and essentially every large-scale hardware model is now based on the topological model of computation. In these models, error-corrected quantum bits (logical qubits) are deﬁned by quantum correlation (entanglement) of numerous zle. These puzzles are what is used by physical quantum bits. Logical qubits carry information, and calculations progress by interacting and entangling these meQuanics. encoded qubits. This process looks like a three-dimensional puzzle, and the same calculation can be done as long as the When quantum circuits are translated topological properties are not changed, which means that it is possible to reduce the volume of the entire graphic by modifying it according to certain rules. into these 3D puzzles, they are unnecessar-

06 Feature │ Virtual Currency Technologies and Challenges ily large. There is a lot of wasted space and were coming from completely different ent during the development and release of therefore wasted hardware resources. Ini- standpoints—some wanting to make the the prototype. Some aspects of the gameʼs tially, Professor Nemoto and her team used tool interesting as a game and some want- interface are still difficult to understand. manual techniques to try and optimise ing to contribute to research. For example, a tutorial or set of instructions these circuits. By using programs such as “The game developers wanted to make about how a puzzle is manipulated is Google SketchUp, her team spent days the tool appealing as a game, while we needed. Also, it was found that players re- painstakingly compressing and shrinking wanted to maintain as much scientific acquire an intuitive game interface that al- topological quantum circuits. This ap- curacy as possible. This led to some heated lows you to sit down and start playing proach was very error prone as small mis- discussions.” without having to think too much. Making takes with manually optimising circuits de- Looking back on the gameʼs develop- the goal of the game interesting and easy stroyed their functionality. The meQuanics ment, Professor Nemoto says, “A sense of to understand is also a key component in game essentially evolved out of the need speed was necessary, and unless the peo- making the platform successful. to develop software that would automati- ple developing the game felt it fascinating, Professor Nemoto is planning to widen cally enforce the quantum mechanical there wouldnʼt be any success.” The con- the range of players by resolving these is- rules. If the software made sure that no centration and excitement of the develop- sues, while also targeting mobile environ- mistakes were made, then you could opti- ment process were key to making the ments. Research activities and game devel- mise freely without constantly checking game fun. opment require different ways of thinking, that trivial mistakes were not appearing. During the development of meQuanics, different timescales, and different ways of As Prof. Nemoto and her team slowly de- a married couple who were working as vis- using funds, and Professor Nemoto is veloped this software, it became clear that iting researchers brought along their two therefore also considering financing it could be turned into a popular game. children, who became absorbed in the through crowdfunding. Both science minded people and casual game and fought with each other to play If a wider range of people start playing gamers could then participate and help it. Seeing that, “I felt confident that it could the game and obtaining high scores, we solve the problem. “I was a complete nov- be exciting enough,” laughs Professor will get closer to realizing a topological ice at developing computer games, but Nemoto. quantum computer. This is the starting when I tried creating a tool that simply al- Gathering a wider point in the first crowd-source attempt to lowed users to modify a circuit according range of wisdom solve an important problem not only in to the rules, I was surprised to find that it quantum information, but in physics itself. started to look like a computer game.”, re- The alpha version of meQuanics was re- (Interview/Report: Akio Hoshi, calls Prof. Nemoto. leased in May 2013. A number of issues Photography: Yusuke Sato) A multinational team that needed to be resolved became appar-

Nevertheless, it took many collaborators and extraordinary eﬀort to reach the point of releasing the tool as a game. The game development project began in January 2013. The development work was done by a multinational team, with no two people from the same country. In addition to Professor Nemoto, the team consisted of specially-appointed Assistant Professor Si- mon Devitt, who is Australian, a French de- signer, a German game programmer/de- signer, and a British back-end developer. Professor Nemoto describes the intense discussions by this team of diverse and strong individual skills and characters as “explosive”. “We would have explosive but exciting discussions at a frequency of about once every week or two” (Professor Nemoto). This was primarily due to the clashes of wills of talented people who

Kae Nemoto

2015 NII Today │ No.56 07 Crowdsensing Useful in Solving Social Problems Using environmental sound maps in tourism and disaster prevention policy making

Masanobu Abe Noboru Sonehara (Professor and Vice-Dean, Faculty of Engineering, (Professor and Director, Information and Society Research Division, Okayama University） National Institute of Informatics/Professor, School of Multidisciplinary Sciences, The Graduate University for Advanced Studies)

Professor Masanobu Abe, Okayama Uni- vironmental sound collection system using researcher in speech processing and had versity, and NIIʼs Professor Noboru Sone- crowdsensing? worked in fields such as lifelogs and re- hara are currently working together to Abe: Suppose, for example, that an adver- gional monitoring systems. build an environmental sound collection tisement for a residential building de- ––And the collection of this environmental system using crowdsourcing and crowd- scribes a neighborhood as a “quiet residen- sound is done using smartphones. sensing. In crowdsensing, smartphones tial area”. There is nothing to show Abe: By using smartphones and having and vehicle-mounted devices used by quantitatively just how quiet the area is. large numbers of people participate, we participants serve as sensors and collect People often find, when they actually live are able to collect information uniformly, data. This joint project aims to recruit in an area, that it can be noisy depending both temporally and spatially, rather than members of the public to collect environ- on the time of day. Or, you might stay at a just having a monthly fixed-point observa- mental sounds from the streets using hotel where the walls are so thin that you tion, for example. The “resolution” is com- their smartphones, and to make use of can hear the sound of the shower and peo- pletely different from that of conventional this data in policies for attracting tourists ple talking in the neighboring room. You measuring methods. and solving social problems, including for can check the visual information of a place Sonehara: Also, crowdsourcing is a method disaster prevention and crime preven- to some extent using Google Street View, that is familiar in tourism development. For tion. for example, but sound information is sur- example, there is a sightseeing spot in Ja- prisingly scarce. That was our starting pan that, for some reason, is only visited by Using the power of the public in point. Thai people. It appears to have become policy making Sonehara: Nowadays, I am often involved popular as a result of Thai visitors to Japan in policy making on regional tourism, di- posting on social networking sites such as ––What made you start developing an en- saster prevention, and so on, and there are Twitter. Crowdsourcing is effective in this requests to make use of sound information way for discovering sightseeing spots. If we in these areas. Osaki City in Miyagi Prefec- could collect a variety of information such ture suffered a great deal of damage from as “beautiful”, “peaceful”, “dirty”, in a simi- the Great East Japan Earthquake, but it is lar way to the “Like” button on Facebook, recovering and crowds are returning. we would be able to easily create sightsee- There is a desire in the city to create a map ing maps that reflect the mental images of of the bustling areas so that people can get the people who have visited. a real sense of the cityʼs liveliness. If you Similarly, this kind of public power can were to visit as a tourist as well, you would be used in policy making. For example, if probably appreciate knowing in advance crowdsourcing could be used to take up which parts of the city are buzzing and comments from city inhabitants, such as which stores are doing well. I spoke to Pro- “The streets are dirty” or “The railing on the fessor Abe about whether it would be pos- bridge is damaged,” it could be useful in sible to collect this kind of sound infor- prioritizing administrative services. More- mation because I knew that he was a over, responses could be based on scientif-

Masanobu Abe

Received masterʼs degree from Waseda University in 1984. Joined Nippon Telegraph and Telephone Public Corporation in same year. Served as Project Manager at NTT Cyber Solutions Laboratories before becoming a professor in the Graduate School of Natural Science and Technology, Okayama University in 2010. PhD (Engineering). 08 Feature │ Virtual Currency Technologies and Challenges Quiet Noisy A notable difference is found 1.2km ically grounded data, rather than instinct Okayama Station Concert Okayama Station Concert and experience as they are now. This kind hall hall of policy making based on citizen participation using IT and other technologies is known as “civic tech”, and nowadays as local government finances are tight, civic tech is showing promise as a trump card for cutting public finances. If environmental City Hall City Hall sound could be added as part of this infor- Approx. noon Approx. 6 pm mation and displayed quantitatively, it Okayama City, November 2014 would have the potential to be useful in various contexts, including for noise con- Figure Visualization of noise levels trol measures, town revitalization, moni- Noise levels measured in Okayama at around noon and 6 pm, November 2014. Colors indicate observation points, that is, places where there were people carrying smartphones containing the developed app. At both times shown, the area toring, safety and security, and disaster around the station is orange, meaning somewhat noisy. However, the area inside the ellipse is noisier at 6 pm. The map makes it very clear which areas of the city are noisy. Also, by using the environmental sounds to estimate the types of prevention. sources that are causing the noisiness, it is possible to extract higher-order information such as places where there are a lot of people and places where there are a lot of cars. Passive and active participation City using a prototype app, and even look- ed value can be created by superimposing ––Specifically, how will the information be ing at the collection results of just four or various data. I think that crowdsourcing collected? five people, it is clear that there are large and crowdsensing are extremely useful Abe: There are two types of environmental differences in the power of environmental techniques in the sense that they compen- sound collection. One type senses the sound depending on location between sate for things that computers cannot do power of sound (decibels). The app we noon and six in the evening. You can see at using people power and make it possible have developed works in the background a glance that the downtown area gets to collect diverse information. Sonehara when participants are using smartphone busy in the evening. : Some say that this kind of re- apps such as Twitter and Line, and it auto- The bene ts of data protection search is not academic, but our efforts are matically measures and uploads the power and utilization in presenting a platform for solving social of ambient sound together with the time problems using raw data. Going forward, and place. In doing so, for the sake of priva- ––Guaranteeing the reliability of data is an our main role will increasingly be to re- cy, the app does not collect information issue in crowdsourcing. spond to the needs of society, and the Abe such as the content of conversations or : I think that this is exactly where power of the public will be a key element in what the sounds are. The other type of col- crowdsensing is useful. Rather than just re- this. lection acquires the actual sound. We are cording data, superimposing physical (Interview/Report: Madoka Tainaka, trying to build an environmental sound da- sounds and location information proves Photography: Kazuyoshi Oyama) tabase by having participants upload that the participant really was at that loca- sounds that leave an impression on them, tion and can improve the reliability of the such as the babbling of a stream or the information. Sonehara shrilling of a cicada. The app provides for : Obviously, it is essential that pri- both passive and active participation. vacy be considered. The challenge is how For sounds collected through active par- to increase the value of data while preserv- ticipation, the app has buttons that allow ing privacy. If benefits such as being useful the participant to subjectively evaluate in regional development or disaster pre- noise level and crowdedness level on a vention become clear, the publicʼs cooper- scale of 1 to 5. This enables us to find that ation can probably be won. Participation the same level of sound is perceived differ- can also be encouraged through incentives ently depending on the time of day and such as discounts or points. Abe location. Sound has a considerable influ- : When big data is mentioned, the focus tends to be on the volume of data, but ence on peopleʼs minds and bodies. People may feel unwell while exposed to noise, in fact, the source of big dataʼs value is in even if they do not think that the sound is the diversity of the information. Unexpect- unpleasant. If we create a noise map that includes subjective evaluations, it may be of use in health care. We have already started collecting data on a trial basis in the center of Okayama Noboru Sonehara

2015 NII Today │ No.56 09 News Five SINET presentations 1 across Japan ̶Increasing interest in 100 Gbps SINET5

NII holds annual presentations on the Sci- ence Information NETwork (SINET), Academic Access Management Federation (GakuNin), and UPKI Digital Certificate Services, in order that this advanced academic information infrastructure—being built and operated by NII—is used and promoted more effectively. This year, NII hosted five presentations across Japan starting in Fukuoka on November 4, followed by Kyoto (Nov 5), Sapporo (Nov 20), Nagoya (Nov 30), and Tokyo (Dec 7). The events included discussions with users and encouraging the circulation of a wide range Wireless Roaming Network), and GakuNin individual consultations. of academic information. In April of next year, Cloud. At the Kyoto and Tokyo venues, the SINET is an information communication the network will migrate to SINET5, which presentations included an update on the network built and operated by NII as academ- will realize an ultrahigh-speed network con- state of development of academic content ic information infrastructure for universities necting the whole of Japan at 100 Gbps. services for SINET5. and research institutions throughout Japan. The presentations at each venue focused There were large numbers of attentive par- The network has nodes (network connection on providing an overview of SINET5, which ticipants at each venue, with around 80 peo- points) nationwide, and it provides universi- will start operation in fiscal 2016, and ex- ple in Fukuoka and 200 in Kyoto (photo- ties and research institutions with an ad- plaining the work of migration to SINET5, as graph). The individual consultations vanced network for supporting the formation well as explaining other infrastructure includ- answered questions about SINET5 migration, of communities among the large numbers of ing the UPKI Digital Certificate Services, Ga- switching to the shared repository service JAI- people involved in education/research and kuNin, eduroam (International Academic RO Cloud, and other topics.

News Exhibition at 17th News Conveying the appeal of 2 Library Fair & Forum 3 research using demos ̶Around 880 visitors, and Forum presentation ̶Inter-University Research Institute Corporation Symposium 2015

The 17th Library Fair & NII exhibited at the Inter-University Research Institute Corpora- Forum was held at Pacifico tion Symposium 2015 held at Akihabara UDX on November 29. Yokohama on November NIIʼs booth presented two demonstrations designed to show 10–12, and NII exhibited a the appeal of NIIʼs research to the general public and particularly booth introducing the lat- to young people who will take on future research. In a demo of est developments in NIIʼs PrivacyVisor, privacy-protecting glasses developed in the labora- services for libraries and li- tory of Professor Isao Echizen (Digital Content and Media Scienc- brary users. es Research Division), visitors who tried on the glasses were able In a presentation titled to experience how they prevent camera face detection. The labo- “An A to Z of NII Services— ratory of Associate Professor Michihiro Koibuchi (Information NIIʼs Developing and Deepening Services” in the forum on the first day, Systems Architecture Science Research Division) displayed a “sub- members of staff in NIIʼs Scholarly and Academic Information Division, merged computer” with a completely waterproof motherboard who are responsible for actually operating these services, presented an that was immersed in a tank of water. The researchers are at- overview of the services and how to use them (photograph). The forum tempting to reduce power consumption by cooling computers was also broadcast via Ustream. using natural energy, and visitors asked Associate Professor Ikki NIIʼs booth introduced services such as CiNii Dissertations, ERDB- Fujiwara (same division), who was explaining the technology, JP, JAIRO Cloud, KAKEN & JST project databases, and NII-REO. about its potential applications. Around 880 people visited the booth, and there was much interest The Symposium also featured “Researcher Talks”, where re- in CiNii Dissertations, a one-stop service for retrieving doctoral dis- searchers spoke about latest topics and the appeal of their re- sertations in Japan that only became fully available to the public in search. From NII, Associate Professor Koibuchi gave a lecture October. titled “Architecture of Future Computers: Using Water and Light”.

▶ Hosting the 2nd SPARC Japan In seminars titled “Science and Research Japan, explained the current status of re- Flash Seminar 2015 Data” and “Research Data Infrastructure of search data use and sharing. NII hosted the 2nd SPARC Japan Japan”, NIIʼs Associate Professor Asanobu Ki- In panel sessions, researchers and librarians Seminar 2015 (Open Access Sum- tamoto (Digital Content and Media Sciences discussed the research support environments mit 2015) on October 21 to coincide with In- Research Division) and Professor Keizo Oya- in demand today. ternational Open Access Week. The keynote ma (Digital Content and Media Sciences Re- address was given by Secretary General of the search Division), as well as other researchers Research Data Alliance (RDA), Mark Parsons. involved in academic information from across

10 News “ Topics Arti cial Intelligence Project Can a Robot Get into the University of Tokyo?” Deviation score of 57.8 in the practice test for National Center Test for University Admissions

The artificial intelligence (AI) project “Can a Robot Get into the University of Tokyo?” (To- dai Robot Project) presented this fiscal yearʼs results on November 14 in Fukutake Hall, In- terfaculty Initiative in Information Studies at Hongo Campus, the University of Tokyo. The Todai Robot Project began in fiscal 2011. University entrance exams are integrated tasks, the results of which can be evaluat- ed quantitatively using points and deviation scores, and the aim of tackling these exams is to present an objective benchmark for AI evo- lution that can act as a guide when considering questions such as “In which fields could AI replace humans?” The goals of the project are to achieve a high score in the National Center Test for University Admissions by 2016 and to pass the entrance examination for the Univer- sity of Tokyo by 2021. The Todai Robot Project is a joint research project in which university and corporate laboratories and research departments take the exam in subjects consistent with their respective research goals. The scores of the organization or team with the best result in each subject are taken as the grades achieved by “Torobo-kun”, the name of the AI being re- searched/developed in this project. This fiscal year, Torobo-kun took the June 2015 practice version of the Center Test run by Benesse Corporation. The AI achieved good results with a total score of 511 (nation- ject, Torobo-kun recorded deviation scores ams in August for entry into the University of al average, 416.4) and a deviation score of above 60 in three subjects: Math I A (devia- Tokyo aimed at essay-type individual achieve- 57.8 in 8 subjects across 5 fields of study. tion score 64.0), Math II B (65.8), and World ment tests (second-stage exams) in geogra- These results give Torobo-kun a chance of at History B (66.5). phy and history (world history) and math (in least 80% of being accepted by 1,055 depart- Also, with the cooperation of Surugadai humanities and science). The AI achieved a ments in 441 private universities and 39 de- Gakuenʼs Sundai Preparatory School, Toro- deviation score of 54.1 in its first attempt at partments in 33 public universities. By sub- bo-kun took the 1st (2015/2016) Practice Ex- the world history test.

“ Hey, this is great!” Assistant Professor Takuya Akiba has ad- SNS Hottest articles on Facebook and Twitter (August–November 2015) vanced into the ﬁnals of the Google Code Jam, a global programming contest being ing PrivacyVisor. PrivacyVisor is the name for held in Seattle. (August 11, 2015) National Institute of Informat- ics, NII (of cial) Facebook the glasses developed by Professor Isao Echi- zen and his team in NII s Digital Content and www.facebook.com/jouhouken/ ʼ Bit on Twitter! Twitter Media Sciences Research Division that protect @NII_Bit Towards commercialization of PrivacyVi- the privacy of the wearer by making it impossi- sor for preventing face detection ble for cameras and other devices to recognize Assistant Professor Akiba, leader of Team Un- Helping to develop local industry by im- their face. (August 06, 2015) agi, is looking very pleased the day after his plementing research ﬁndings in society team was announced the winner of the ICFP Hyakuo Makino, the Mayor of the Sabae City Programming Contest 2015! Congratula- National Institute of Informat- in Fukui Prefecture—a city famed for making ics, NII (of cial) Twitter tions! (September 02, 2015) eyewear—announced today (August 6) that @jouhouken *Some text omitted. companies in the city would be commercializ-

2015 NII Today │ No.56 11 Letʼs look at the potential of crowdsourcing from the perspective of the problem of unemployment. It is important for the nation that unemployed people find work. Con- Essay sidering crowdsourcing websites as labor markets where employers seeking workers and people seeking jobs are matched together, is it possible to solve the problem of unemployment by analyzing the activity logs of these employers and people seeking jobs? Can In economics, the following reasons are given for there being unemployed people in Japan who would like to work: ❶ there are job vacancies, but the unemployed person is Crowdsourcing incapable of searching for them (in job magazines, etc.); ❷the employer with job vacan- Solve the cies to fill cannot find the unemployed people; ❸there are no job vacancies (in the unemployed personʼs area of residence); ❹ there are job vacancies, but the unemployed per- Problem of son does not have the required skills; and ❺ there are job vacancies, but the pay is low and so the unemployed person chooses to be unemployed while looking for another job. Unemployment? It is anticipated that Reasons ❶, ❷, and ❸ could be addressed by crowdsourcing that makes all information about employers seeking workers and people seeking jobs easily searchable, and also makes it possible for people to work without being dependent on a particular area. With regard to Reason ❹, small-to-medium-sized companies find it difficult to recruit workers despite not requiring high-level skills, and so it may be the case that the unemployed person is looking only for jobs in large companies without realizing that jobs matching their skills are only available in small-to-medium-sized companies. Reason ❺ may be because employers seeking workers do not realize that the pay they are offering is lower than the market rate. In any case, the economics perspective is that with crowdsourcing sites where everyone can obtain large amounts of information, there will be fewer unemployed people. Then is it really sufficient to simply provide both employers seeking workers and people seeking jobs with lots of information? In some cases, there may be information that is obscure or hinders matching, and so not only will the number of unemployed not de- crease but it may actually increase. For example, letʼs consider a site that provides a rating history of employees by employers and a rating history of employers by employees. Job seekers will flood highly rated employers with applications, and employers seeking workers will compete over highly rated job seekers. Jobs (or people) will become concentrat- ed on the highly rated job seekers (or employers), and their ratings will successively rise. Meanwhile, those who do not get a good rating at the outset will not receive any jobs (or people) at all, and so they will not be able to improve their rating. In the end, without matching progressing smoothly, there will still be many unemployed people and companies with staff shortages. Unlike offline labor markets, crowdsourcing sites record activity logs. Could these logs be used to identify how much information users can process and what is obscure or un- Takayuki Mizuno necessary information that hinders matching? If, based on this knowledge, a system Associate Professor, Information and Society could be created for conveying the necessary information in simplified form, it could also Research Division, National Institute of Infor - prove useful in the provision of offline information to job seekers and employers at, for matics example, HelloWork, a public employment service center.

Future Schedule January 12│The 9th Science Council of Japan Symposium on Content and Media Sciences Research Division). Date will be Informatics (Sponsor). For details and registration, go to the announced on NIIʼs official website (http://www.nii.ac.jp/). event information page at (http://www-higashi.ist.osaka-u. February 25│Public Lecture No. 6 “Your Information, Whose Is ac.jp/scj/symposium09.html). It?—Personal Information and Privacy in the Age of Big Data” January 19│The 3rd SPARC Japan Seminar 2015 “The Potential (Specially Appointed Lecturer Rihoko Kawai, Information of Social Media Services for Researchers” (provisional title, Systems Architecture Science Research Division). For details and Host). registration, go to NIIʼs official website (http://www.nii.ac.jp/ January or February │ The 5th Industry-Government- event/shimin/). Public Lecture No. 5 (January 27) canceled at Academia Collaboration Lecture, “The Development of host's discretion. Research on Material Perception” (Professor Imari Sato, Digital March 3 & 4│Cyber Security Symposium in Dogo (Exhibitor).

Notes on cover When the comments and actions of many robots are shared via a network and analyzed, they become valuable information. The illustration illustration shows this exciting crowdsourcing scenario.

Search “NII Today”! Weaving Information National Institute of Informatics News [NII Today] into Knowledge No. 56 Dec. 2015 [This English language edition NII Today corresponds to No. 70 of the Japanese edition.] Published by National Institute of Informatics, Research Organization of Information and Systems Address│National Center of Sciences 2-1-2 Hitotsubashi, Chiyoda-ku, Tokyo 101-8430 Publisher│Masaru Kitsuregawa Editorial Supervisor│Ichiro Satoh Cover illustration│Toshiya Shirotani Copy Editor│Madoka Tainaka 12 Feature │ CrowdsourcingProduction│ WillMATZDA Change OFFICE Society CO., LTD., Athena Brains Inc. Bit (NII Character) Contact│Publicity Team, Planning Division, General Aﬀairs Department TEL│+81-3-4212-2164 FAX│+81-3-4212-2150 E-mail│[email protected] http://www.nii.ac.jp/en/about/publications/today/