SNS活用の分析におけるデーターサイエンスの活用
Summary
TLDRこのプレゼンテーションでは、筑波大学のMuneo Kaigo教授が、データサイエンスを活用して市民社会と自治体がソーシャルメディアを利用する状況を分析する研究について語ります。東日本大震災後のソーシャルメディアの役割や、自治体と市民グループ間のコミュニケーションの改善、そしてデータサイエンス手法である指数平滑法を用いた定期的なコミュニティ活動低下の分析について紹介しています。彼の研究は、市民社会の活性化と自治体の柔軟性向上に大きな影響を与える可能性を示唆しています。
Takeaways
- 🎓 スクリプトは筑波大学人文社会科学部の教授であり、海外のInstitut d'études Politiques de Bordeauxの客員教授である氣賀雄朗氏によるプレゼンテーションです。
- 🔍 氣賀氏は、1990年代にテレビニュースの影響や逆影響に関する研究を行っており、インターネットの普及に伴う情報格差やネットワーク社会の問題に注目しました。
- 📈 2011年の東日本大震災後には、社会メディアと災害に関する研究を開始し、地域政府と市民圏のコミュニケーションを促進するプロジェクトに取り組みました。
- 🗨️ 研究では、地域政府が社会メディアを活用し、市民グループや市民間のコミュニケーションを促進することの重要性が強調されています。
- 📊 データサイエンスは、社会メディアの活用状況や市民圏の情報格差を分析する上で非常に有用であることが示されています。
- 📉 社会メディアでのコミュニティ活動は、適切なケアがなければすぐに非活発化する傾向があることが学ばれました。
- 📈 氣賀氏は、Facebook広告の有効性や顔を合わせる機会を作り出すことでコミュニティの活性化が促進できることを発見しました。
- 📚 2017年に出版された書籍「Social Media and Civil Society in Japan」は、このプロジェクトの基礎を形成しており、興味のある読者は参考にできるでしょう。
- 📉 データ分析によると、3月から5月にかけてコミュニティの活動レベルが定期的に低下する周期的な現象が確認できました。
- 📊 「指数平滑法」というデータサイエンスの手法を用いて、時間序列データを分析し、周期的な低下の証拠を立証しました。
- 🔑 プロジェクトの成果として、地域政府と市民グループの間の相互作用が増加し、コミュニティの統合が進み、相互理解が深まりました。
Q & A
どのようなテーマについてのプレゼンテーションを行いましたか?
-データサイエンスを利用して市民社会を分析し、地方自治体がソーシャルメディアをどのように使用するかについてのプレゼンテーションを行いました。
Muneo Kaigoさんはどのような経歴を持っていますか?
-筑波大学人文社会科学部の教授であり、かつボルドーの政治学研究所の客員教授でもあります。また、コロンビア大学のテレインフォメーション研究所で訪問学者として過ごした経験もあります。
Muneo Kaigoさんの過去の研究はどのようなものでしたか?
-1990年代にはテレビニュースの影響と逆影響に関する研究を行い、インターネットの普及に伴う情報格差やネットワーク社会の問題にも取り組みました。また、双方向ビデオ通信やファイル共有、さらには核リスクに関する情報にも研究を進めていました。
東日本大震災後にどのような研究を始めましたか?
-東日本大震災後には、ソーシャルメディアと災害に関する研究を始めました。この研究は、地方自治体と市民団体の間、市民団体同士の間のコミュニケーション活動におけるソーシャルメディアの使用を分析するものでした。
「市民参加ポータル 筑波市民のための」Facebookページとは何ですか?
-筑波市の市民参加部門によって運営されているFacebookコミュニティページで、現在も活動しています。このページは、市民とのコミュニケーションを促すために設立されました。
Muneo Kaigoさんの研究プロジェクトの目的は何でしたか?
-研究プロジェクトの目的は、ソーシャルメディアを通じて地方自治体と市民団体の間、市民団体同士の間のコミュニケーションを促進し、情報格差を解消することです。
プロジェクトの実施中にどのような問題を発見しましたか?
-プロジェクトの実施中に、ソーシャルメディアでの活動レベルが3月から5月にかけて定期的に低下する周期的な現象を発見しました。また、広告を使用することでコミュニティの成長を促すことができる一方で、既存のコアメンバーの活動が低下するという副作用も見つかりました。
データサイエンス手法をどのように使用して問題を解決しましたか?
-データサイエンス手法の1つである指数平滑法を用いて、活動レベルの低下が周期的に再発する証拠を見つけ、問題を解決しました。また、広告や対面での交流の機会を増やすことで、コミュニティの活動を促進しました。
プロジェクトの結果としてどのような影響が見られましたか?
-プロジェクトの結果、筑波市の職員と市民団体の間の相互作用が急速に増加し、互いの理解が深まり、コミュニケーションが改善されました。また、市民団体の不満やストレスも軽減されました。
今後の研究でどのような課題を検討予定ですか?
-今後の研究では、オンラインコミュニティが自己維持できるために必要な最小会員数を検証する予定です。また、データサイエンス手法をさらに活用して、より効果的な分析方法を探求することも検討しています。
Outlines
😀 開会致辞と自己紹介
講演者は、筑波大学人文社会科学部の教授である榊浩朗であり、また、海外のInstitut d'études Politiques de Bordeauxの客員教授でもあります。コロンビア大学のColumbia Institute for Tele-Informationで客員研究員として過ごした経験も持っています。現在、アジアメディア情報通信センターの代表として活動しており、過去にはテレビニュースの影響に関する研究を行い、PhD論文を完成させています。インターネットの普及に伴い、情報格差に関する研究を進め、その後はブロードバンドの利用拡大に伴う双方向ビデオ通信やファイル共有の問題に注目しました。また、原子力リスクに関する研究も行っており、東日本大震災後には、ソーシャルメディアと災害に関する研究を開始しました。現在は、政治的、社会的不平等に関する研究に取り組んでおり、今回の講演では、地方自治体がソーシャルメディアを活用し、市民団体とコミュニケーションを取る方法と、データサイエンスがその分析にどのように役立つかを紹介します。
🏙️ 地域コミュニティの役割とソーシャルメディアの活用
榊は、地域コミュニティがルールの遵守やコミュニティ保護のために重要な役割を果たすことを強調しています。地域のボランティアが地方自治体と密接に協力することが必要です。市民団体から話を聞くことで、彼らは現状に不満があり、特に若い世代との交流を望んでいることがわかりました。彼らは異なる市民団体間の交流も増やしたいと考えています。東日本大震災時には、Twitterを通じて市役所職員と市民の間で情報交換が活発に行われ、その後もSNSを通じての情報交換が続きました。榊は、この経験から「復興力」の重要性を学び、ソーシャルメディアの活用がその実現に役立つと認識しました。
📈 ソーシャルメディアのコミュニティ形成と広告の効果
榊は、ソーシャルメディアでの「ウイルスマーケティング」が誤解を招く可能性があることを指摘し、実際には、ソーシャルメディアを通じて情報を迅速に拡散することは難しいと認識しました。しかしながら、悪質な噂やスキャンダル、そしてフェイクニュースは迅速に拡散されることがあることを強調しています。彼は、コミュニティの活性化のために広告の使用や、顔を合わせる機会を作り出すことが重要であると学びました。また、広告を利用することでコミュニティを拡大しようとすると、既存のコアメンバーの活動度が下がるという意図しない効果があることも発見しました。
📊 データサイエンスの有効性とコミュニティの活性化
榊は、データサイエンスの方法を用いて、コミュニティの活動レベルの低下を分析し、その原因を特定しました。指数平滑法を使用して時間序列データを分析し、3月から5月にかけて活動レベルが低下する周期的な現象を指摘しました。この方法を用いて、ノイズを除去し、データの変動を理解することができました。それにより、問題に対処するための対策を立案し、2015年までに問題を解決しました。プロジェクトを通じて、オンラインコミュニティは2,000人以上のメンバーを持ち、活動が活発になり、市役所職員と市民団体の間の相互作用が増加しました。
📚 プロジェクトの成果とデータサイエンスの重要性
榊は、プロジェクトを通じて発見された問題と、それを解決するために使用したデータサイエンスの方法を共有しました。スプレッドシートの限界を超える場合、データサイエンスの方法は非常に価値があります。彼は、データサイエンスを学び活用することを皆さんに勧め、プロジェクトの詳細や成果を紹介しました。また、コミュニティが自己完結するにあたり、最小限のメンバー数が必要であるという仮説を立て、その具体的な数値についても考察しています。
Mindmap
Keywords
💡データサイエンス
💡市民社会
💡ソーシャルメディア
💡情報_GAP
💡指数平滑
💡地域政府
💡コミュニティ活動
💡災害対応
💡仮想コミュニティ
💡自己維持可能性
Highlights
Muneo Kaigo教授介绍了自己在数据科学和社交媒体分析领域的研究背景。
Kaigo教授在1990年代研究了电视新闻的影响及其反向效应,并以此作为其1999年博士论文的基础。
随着互联网的普及,Kaigo教授开始研究由此产生的信息差距和网络社会面临的问题。
Kaigo教授还研究了双向视频通信和文件共享问题,这些问题源于宽带的日益普及。
Kaigo教授对核风险相关的信息进行了研究,并提出了一些潜在问题,但这些问题当时并未受到重视。
东日本大地震后,Kaigo教授开始研究社交媒体和灾害之间的关系。
Kaigo教授目前参与的研究涉及政治不平等、社会不平等和差距。
Kaigo教授介绍了他如何通过分析社交媒体的使用来研究地方政府与民间团体之间的沟通活动。
Kaigo教授强调了数据科学研究方法在分析社交媒体和民间社会信息差距中的有用性。
Kaigo教授提到了他在2010年代进行的研究,特别是东日本大地震是他开始这个项目的主要原因。
Kaigo教授介绍了他如何使用Facebook社区页面来促进地方政府与居民之间的互动。
Kaigo教授讨论了地方政府和民间社会在资源和人力上的不足,以及民间社会在国家中的角色。
Kaigo教授指出,民间团体和居民协会的成员对现状不满,并希望与更多居民,特别是年轻人进行更多互动。
Kaigo教授分享了他在东日本大地震期间如何通过Twitter与居民交换信息的经验。
Kaigo教授强调了社交媒体在增强社区抗灾能力方面的重要性。
Kaigo教授介绍了他如何通过数据科学方法来分析社区活动的季节性下降趋势。
Kaigo教授讨论了使用广告来增加Facebook社区页面成员和活动的有效性。
Kaigo教授指出,广告的使用对社区成员的活动有正面和负面的影响。
Kaigo教授分享了他如何通过数据科学方法解决社区活动下降的问题,并在2015年取得成功。
Kaigo教授提出了一个假设,即在线社区需要大约2000名成员才能自我维持。
Transcripts
Good afternoon. Today, I would like to give a presentation on the topic of utilizing data science to analyze civil society
and the use of social media by local governments.
I am Muneo Kaigo from the Faculty of Humanities and Social Sciences, University of Tsukuba,
and I am pleased to have this opportunity to talk to all of you.
I would like to begin with a brief introduction of myself. I am a professor at the Faculty of Humanities and Social Sciences, University of Tsukuba,
and a visiting professor at an overseas university called Institut d'études Politiques de Bordeaux.
From 2004 to 2005, I spent a year as a visiting scholar at the Columbia Institute for Tele-Information at Columbia University.
I am currently serving as a representative of Japan at the Asian Media Information and Communication Centre, an international NGO.
I would like to say a few words on my past research. In the 1990s,
I had conducted research on the effects and inverse effects of television news, which became the basis of my PhD dissertation that was approved in 1999.
Around that time, the use of the Internet started gaining traction among the general public,
which prompted me to conduct research on the resulting information gap
and issues confronting a network society.
To further deepen the scope of my research, I was working on problems related to two-way video communication
and file sharing that stemmed from the growing utilization of broadband.
During the same period, I was also conducting research on information related to nuclear risks.
I had raised several potential issues at the time, but unfortunately, my concerns were not taken seriously.
The Great East Japan Earthquake then happened.
After that, I began to conduct research on social media and disasters, which is what I would like to talk about today.
Currently, I am involved in research on political inequality,
as well as social inequality and disparity, with professors
from the University of Tsukuba and many other universities.
I would like to begin by briefly explaining the topic of my presentation today.
I hope to share with you what I have found from analyzing research activities that have encouraged the use of social media by local governments,
as well as its use in the communication activities between the local government and civic groups, and between different civic groups.
In particular, I would like to explain how data science research methods can be useful in this regard.
This was a research project that I had completed some time ago,
and my focus today is on the information gap between social media and civil society,
a problem unique to the Internet environment in Japan.
I am still following up closely on the results of this research as a self-sustaining project.
Although I have mentioned that this research was carried out in the past,
it was actually conducted in the 2010s, which was still relatively recent.
In fact, the Great East Japan Earthquake was the main reason why I began working on this project.
A number of collaborative projects involving the University of Tsukuba, Tsukuba City,
and Intel Corporation were launched at the time,
and I had used the findings from those projects to conduct this research.
The Facebook community page called “Civic Engagement Portal for the Residents of Tsukuba City” remains active today.
My research on this topic formed the basis for Social Media and Civil Society in Japan,
a book I published in 2017.
Please take a look at this book as well if you are interested.
There are some important points to know regarding the background of my research, which I will now briefly discuss.
Very unfortunately, it remains a fact that our national government
and most local governments lack sufficient funding and manpower.
This has made it impossible for them to cover all our needs.
Civil society plays an especially vital role in situations such as the one in our country.
Typically, the phrase “civil society” tends to evoke a rather foreboding and scary image.
Many people also think of residents’ associations, neighborhood associations,
and ward associations as a source of annoyance and would rather avoid them if possible.
Perhaps many of you and your family members feel the same.
I think most people in the world would try to avoid getting involved in these organizations if possible.
However,
it is actually important for the community to play a part in ensuring that the rules established by the local government
—for instance, rules regarding how garbage should be disposed of and the kind of rules necessary for protecting the community—are adhered to.
Things usually do not go very well unless volunteers
in the community work closely with the local government.
As I started listening to what people from civic groups had to say,
it became clear to me that the members of these civil groups and residents’ associations are not satisfied with the status quo.
They hope to have more interaction with fellow residents,
especially with young people like you. I also realized that they want to see more interaction between different civic groups.
This is when we started to explore various ways to make that possible.
This was the premise on which the objectives of this research were built.
When the Great East Japan Earthquake struck, the city hall of Tsukuba City,
where I was living, suffered significant damage from the disaster that made it impossible to serve its intended functions.
Around that time, Twitter became an active channel through
which the city hall staff exchanged information with the residents of Tsukuba City.
Several days after the earthquake,
people were still exchanging information with one another, especially through social media platforms such as Twitter.
Subsequently, I had the chance to talk to other people who were involved in this at the time,
and we came up with many ideas on how we could make local governments function even better during a disaster.
Put a little differently, we were interested in the concept of “resilience,”
and this episode taught us the importance of enhancing our resilience.
At the time, I realized that social media was actually very useful for this goal,
something I’m sure many of you have also realized.
Our goal was to build a community that is resilient to disasters.
I believe that a good way to achieve this would be for local governments
and civic organizations to connect with one another on social media moving forward,
which led to the launch of this project.
Through this series of events, I started serving a consultant-like role for Tsukuba City.
Everything sounds great up till this point, but to be honest,
I had quite a lot of trouble figuring out how to introduce and implement this project as I started thinking about its objectives.
No matter how much research I did on related subjects,
I could not find any good precedents.
I also did not find any successful initiatives in the past that I could draw on.
I had a hard time identifying relevant initiatives that could be used as a reference,
especially since the sociocultural conditions of Japan and other countries were so different.
I was also very troubled to find out that the methodology of the research had not been established at the time.
I decided to come up with a short-term strategy as Tsukuba City wanted me to get down to work as soon as possible,
even though I had not thought of any sustainable solution.
Tsukuba City wanted to stick to a formalist approach for the project,
and we decided to proceed with it even when it felt as if we were groping in the dark at times.
Next, I will give you an outline of the experimental study for this research, which began in February 2012.
As I mentioned earlier, the “Civic Engagement Portal for the Residents of Tsukuba City”
Facebook page run by the Civic Engagement Division of Tsukuba City is still active today.
When I made this slide, the size of this community as measured
by the number of people who have clicked “Like” was 3,069.
One thing I quickly realized after launching the project was that
the “viral marketing” on social media that I had been told about and
read about in various popular books and business books at the time was actually misleading.
I was told at the time that if someone shares something on social media,
the information will travel quickly, making it possible for anyone to share things easily
and disseminate various kinds of information efficiently.
Perhaps this might be true with pictures of cats, dogs, or food,
but the idea that using social media for just about anything would make it successful is simply ridiculous.
I’m sure all of you know this.
However, things like malicious rumors and scandals, and especially fake news these days,
have the power to spread quickly.
It became clear to me that such information can indeed become viral.
Let’s return to our topic.
While virtual communities centered on casual
and appealing subjects can continue to function even without much effort put into them,
communities for civic engagement or volunteer work, such as the one that was launched for the purpose of this project,
will quickly become inactive if they are not properly taken care of.
This may sound a little odd, but it felt as if the community was on life support for a while,
and I had no idea when it would cease to exist.
Before I get to the main subject of my presentation, I want to first explain what I have learned from this project.
One thing I discovered from conducting this research is how effective advertisements are,
especially when you want to increase the number of people following your Facebook community page,
the number of members in your community, or get more people to take a look at your community page.
This includes the use of advertisements in Facebook.
I also wanted people to be active and stay engaged in the right community.
I use the word “action” today to describe this,
but I also wanted people to leave comments on posts
or click “Like” on a comment that someone else had written
as a means of increasing the level of activity in the community.
To do that, people needed to meet each other. In other words,
I realized that it is better to have some face-to-face interaction.
Therefore, I learned the importance of encouraging more activity in the community by placing advertisements
or by creating more opportunities for face-to-face interaction in some cases, and by combining both approaches in other cases.
Something else related to this that I discovered during the project’s implementation
when I was not utilizing advertisements was
how much time it takes to grow an online community in its early stages.
Although I initially started out with advertisements,
I stopped using them after a while, which made the community’s rate of growth very slow.
Without a coherent strategy, I noticed that the level of activity in the community declined significantly.
However, I also found
that using advertisements to expand the community had some unintended effects.
One such effect is that members who had been actively engaged in the community up to that point,
or the so-called core members of the community, started to become less active.
This project has shown that using advertisements to boost community pages
on social media platforms can have both positive and negative effects.
With the project outline out of the way,
I would now like to turn to the main topic of today’s presentation, data science.
I want to discuss how and when data science has been useful.
Let me begin by sharing with you a problem I had encountered in the course of this project.
This was a problem that involved a specific period of the year.
In particular, the data showed some signs
that the level of activity in the community declines from March to May.
The first time this phenomenon occurred was in 2013,
when I was first able to verify it.
In other words, it was the year after the project was launched.
It would be fine if this only happened once, but the same trend occurred again the following year in 2014.
In other words, I noticed a phenomenon where activity levels would gradually decline from March to May on a regular basis.
Well, I had to find evidence of whether
this was in fact a periodically recurring phenomenon.
To do so, I first attempted to map out the activity levels over the entire year.
This is probably the best we can do in terms of visualization that a spreadsheet software is capable of.
The blue line in this figure shows the activity levels in 2013, and the green line shows the activity levels in 2014.
You can see a blue circle in the activity levels,
which indicates the period during which the activity levels are declining.
I first pointed out that there was a decline by marking it in this way.
This may be good enough when I am doing a verbal explanation,
but if I were to present this in my research, there would be nothing I can say if someone asks, “What if it just looks that way to you?”
This is why I hope all of you can learn more about data science moving forward.
In the event that there is something you need to find evidence for in your research data
so that you can make a prediction that maybe something
that has happened before will happen again in the future, data science will be essential.
I argued that there was a periodically recurring phenomenon using this method.
This method is called exponential smoothing of time series data,
which simply means that the data has been assigned weights that decrease exponentially from the newest to the oldest data.
Put differently, this is a prediction method where the older the data is,
the less weight is given to it in terms of the level of priority assigned to the data.
The more recent the data is,
the more relevant it is deemed to be and the greater the weight that is assigned to it.
This method is quite commonly used in short-term forecasting, especially in economics.
The advantage of exponential smoothing is
that it allows noise to be eliminated from the visualized data.
As in our case, this method is useful for making sense of data with quite a bit of fluctuations,
like the data in the previous figure.
That’s why I have tried using it here. As you can see in this slide,
the arrows for 2013 and 2014 appeared after using this method.
The data visualization and analysis indicate a downward trend as shown by this extra arrow,
which allowed me to argue that the decline in activity levels recurs periodically.
By using a method from data science,
I was able to analyze and visualize the problem I had uncovered.
I then developed and put in place some countermeasures to improve the situation over the course of this project,
and the problem was resolved by 2015.
I was able to come up with a solution to this problem
after laying the groundwork for tackling the problem in a so-called physical setting in the real world.
Over time, the online community eventually grew to over 2,000 members
as the project went on, and fortunately, the community became more and more active.
As a result, the number of interactions between the staff of Tsukuba City and civic groups in the city increased rapidly,
even though there was hardly any interaction between them in 2010.
In other words, positive interactions became more frequent,
and there was better mutual understanding between the different parties.
The frustration experienced by civic groups and the staff has also been relieved.
This facilitated better communication between different civic groups,
which made it possible for the different groups to start working together.
This kind of integration between various civic groups has also been echoed within the new online community,
where interesting interactions have been mediated through a variety of events.
Nonetheless, this project is still a case study,
so there are limits to how much its findings can be generalized.
I believe that there are perhaps even better data analysis methods out there.
There are problems that are unique to specific local governments in Japan,
and of course, some problems are unique to certain communities or factors at play.
This makes it difficult to generalize the findings of this project,
although I believe that the results of this research are
still useful for pointing out to other communities that certain things can be achieved.
To bring this research into its next phase,
I have been thinking about what is necessary for this virtual community to become a self-sustaining entity.
One thing I had drawn from this research in the 2010s was the hypothesis
that a minimum number of members is required in an online community for it to be self-sustaining.
As for the specific number of members required,
I looked at the number of people in both the Twitter community and the Tsukuba Facebook community during the earthquake,
which led to my hypothesis that perhaps an online community can be self-sustaining if it had around 2,000 members.
This is a hypothesis that I hope to test in the future.
The main point of my presentation today was to share with you some of the details of my project.
I have described a problem that I discovered in the course of my project,
and how I used exponential smoothing to find the evidence necessary to address this problem.
Spreadsheets are of course useful,
but they also have their limitations.
In such cases, methods from data science can be very valuable.
I would like to encourage all of you to study data science and make the most of it.
Thank you very much for your attention.
Browse More Related Video
【ベストセラー】橘玲「裏道を行け ディストピア世界をHACKする」を世界一わかりやすく要約してみた【本要約】
【Z世代は社会の写し鏡】日本に蔓延する不安ビジネスの正体/社会はZ世代化している/ラインは重い/友達と違ってたら嫌/いい感じに怒ってほしい/消えるブラック、消えない不安/テーマパーク化する大学
【みんな詰んでる】岸田政権支持率毎日17%!共同・総裁再選望む10.4%!政治活動費改革不十分は90.4%!どうみても岸田の政治家人生詰んでる!いや国民の生活はもっと詰んでる!
【要約】ATTENTION SPAN(アテンション・スパン) デジタル時代の「集中力」の科学【グロリア・マーク】
注目AIニュース9選~iPhoneにChatGPT搭載!?、OpenAIの新発表、XのGrokいよいよリリース
【※教授が警告!】マスクとパフォーマンスに与えるヤバい影響|一橋大学准教授が語る|ビックデータと最新研究
5.0 / 5 (0 votes)