=Paper=
{{Paper
|id=Vol-3276/SSS-22_FinalPaper_27
|storemode=property
|title=Feature Concepts as Pattern Language for
Data-Federative Innovations
|pdfUrl=https://ceur-ws.org/Vol-3276/SSS-22_FinalPaper_27.pdf
|volume=Vol-3276
|authors=Yukio Ohsawa,Sae Kondo,Teruaki
Hayashi
}}
==Feature Concepts as Pattern Language for
Data-Federative Innovations==
https://ceur-ws.org/Vol-3276/SSS-22_FinalPaper_27.pdf
Feature Concepts as Pattern Language for Data-Federative Innovations
Yukio Ohsawa1, Sae Kondo2, and Teruaki Hayashi1,
1The University of Tokyo, 2Mie University
1 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan,
ohsawa@sys.t.u-tokyo.ac.jp
Abstract unsupervised machine learning methods, such as clustering
To ensure that all papers in the publication have a uniform with cutting noise events (e.g. [Fränti and Yang 2018]) are
appearance, the authors must adhere to the following instruc- algorithms in which a hidden cluster is restored from data
tions: Feature concepts, an essential tool for data-federative including scattered noise signals. Thus, embedded clusters,
innovation processes, are introduced here as a language to ex-
as the desired information to be acquired from the data, can
press the model of knowledge to be acquired from data. A
feature concept can be represented by a simple feature, such be interpreted as a feature concept, as shown in Fig.1. In ad-
as a single variable, or by a conceptual illustration of the ab- dition, the decision tree [Quinlan 86] realizes a tree as a fea-
stract information obtained from the data. Useful feature con- ture. Feature concepts, if represented explicitly via the com-
cepts for satisfying the latent or explicit requirements in so- munication of participants in the data market, play the role
ciety, or the market of data, are found to have been elicited
of bridging social requirements and features in datasets, as
so far via creative communication among stakeholders. Here,
the contribution of feature concepts to useful findings is illustrated in Fig. 2.
shown with a couple of use cases, for example, explanation
of change in markets and earthquakes.
Introduction
The necessity to elicit information about the data-use con-
texts, that is, the situations where to use data and/or receive
the services or products created based on data, has been po-
sitioned as a key scope in creating a solution for satisfying a
requirement in businesses. Although participants enjoyed Fig. 1. Examples of feature concepts for three basic methods
workplaces for innovations using/reusing data [Ohsawa et for data mining (left: from [Ohsawa 2018b]).
al. 2013], a marketplace for data-federative innovation)
have been urged to speak out requirements and ideas for
their satisfaction so that they can add or revise the DJs and
store the used ideas in the background database, the missing
links between data and the requirement cannot be covered.
To cope with this problem, in this study, a method is intro-
duced to illustrate the abstract image of the information to
be acquired using datasets for requirement satisfaction.
Feature concepts
A feature concept is an abstract image of the information or
knowledge to be acquired using data linked to the method,
that is, how, why, and the dataset(s), that is, what, should be
used to satisfy a requirement. In the examples shown below,
we discover that human creativity in data utilization has
been enhanced by eliciting, using, and sharing concepts in
various forms. These concepts, if the creator explicitly rep-
Fig. 2. The images and positions of feature concepts (FC#) in the
resents, are regarded as feature concepts. Below, we con-
communication to connect requirements to solutions and DJs.
sider the feature concepts illustrated in Fig.1. For example,
___________________________________
In T. Kido, K. Takadama (Eds.), Proceedings of the AAAI 2022 Spring Symposium
“How Fair is Fair? Achieving Wellbeing AI”, Stanford University, Palo Alto, California,
USA, March 21–23, 2022. Copyright © 2022 for this paper by its authors. Use permitted
under Creative Commons License Attribution 4.0 International (CC BY 4.0).
96
Examples of feature concepts in data utilities Feature concepts and pattern language
An example is the change explanation in businesses and sci- Feature concepts may be regarded as a customized pattern
ences, which was elicited as a requirement for supermarkets. language, initially proposed in urban planning [Alexander et
In comparison with the detection or prediction of changes al 1977] and diverted so far to other systems design. Here, a
using machine learning technologies (e.g., [Fearnhead and set of patterns composed of urban elements, such as parks,
Liu 2007, Miyaguchi and Yamanishi 2017]), change expla- bridges, houses, etc. were used to explain and design struc-
nation means linking the observed change in the data to hu- tures of urban areas. Each pattern with an illustration is
man understanding of the dynamics in the real world. Thus, linked to a context, problem, and solution to the problem.
it is essential to create a feature concept for enabling data Individual thoughts and communication toward consensus
visualization that inspires humans to understand the under- within a team engaged in a task of design or other collabo-
lying dynamics. Borrowing the idea of diversity shift pro- rations can be smoothed by using the patterns as a common
posed by Kahn [Kahn 1995] in a market, we drew an image language for expressing contexts, problems, and solutions.
corresponding to the feature concept in Fig.3 and invented Furthermore, the patterns can be connected via relationships
graph-based entropy (GBE [Ohsawa 2018a]) which is an in- from/to each other, which may be hierarchical relations or
dex of the diversity of events on their distribution to the clus- likeliness to be combined. Similarly, once a feature concept
ters in the co-occurrence graph of items in the market. The is created and shared with others, it becomes a tool for inno-
change in GBE is a sign of structural change in the target vators who think and communicate to a federate and/or use
real world and is informative in explaining changes if cou- data. In addition, the relationships among feature concepts
pled with the graph shown in Fig.3, where the bridging edge can be, similarly to patterns in a pattern language, the hier-
between the two clusters is cut in the 10th week of the year, archical structure (e.g., “diversity shift” over “diversity,”
which is interpreted as the growth of the lower cluster cor- etc.), the connectivity (e.g., diversity shift can be connected
responding to spices for cooking stew. The 10th week in the with clusters or with networks), etc. Thus, the links between
data was a hot period in August, but the frequency of the feature concepts or from feature concepts to the real-world
query “stew” in Google increased from August in Japan should come from the communication between data scien-
every year. tists or data scientists with others.
The author then diverted the diversity shift to an analysis of
earthquakes [Ohsawa 2018b]. Here, a model was introduced
to explain the dynamics of earthquakes in two phases: (1) Acknowledgement
the increase in the diversity of epicenter clusters, and (2) the This study is partially supported by JSPS 20K20482
coupling of the clusters due to new activity in the seismic
gap, followed by a large one. The entropy defined in the dis-
tribution of the epicenters increases in phase (1) and de- References
creases in phase (2). Thus, the FC diversity shift used for Alexander,C., Ishikawa, S., Silverstein, M. 1977. A Pattern Lan-
marketing was reused to explain the earthquake precursors. guage: Towns, Buildings, Construction. Oxford Univ. Press, USA.
Fearnhead, P, Liu, Z. 2007. Online Inference for Multiple Change-
point Problems. J. Royal Statistical Soc. B69(4) ISSN 1369-7412
Fränti P., Yang J. 2018. Medoid-Shift for Noise Removal to Im-
prove Clustering., In: Rutkowski L., et al (eds), Artificial Intelli-
gence and Soft Computing, LNCS10841. Springer
Kahn, B.K. 1995. Consumer variety seeking among goods and ser-
vice, J. Retailing and Consumer Services 2, 139-148
Miyaguchi, K., and Yamanishi, K. 2017. Online detection of con-
tinuous changes in stochastic processes, Int J. Data Science and
Analytics 3 (3), 213–229
Ohsawa, Y. , Kido, H., Hayashi, T., Liu, C. 2013. Data Jackets for
Synthesizing Values in the Market of Data, Procedia Computer
Science 22, 709-716, doi.org/10.1016/j.procs.2013.09.152
Ohsawa, Y. 2018a. Graph-Based Entropy for Detecting Explana-
tory Signs of Changes in Market. Rev Socionetwork Strat 12, 183–
203 (2018). https://doi.org/10.1007/s12626-018-0023-8
Ohsawa, Y. 2018b. Regional Seismic Information Entropy for De-
tecting Earthquake Activation Precursors, Entropy 20(11), 861.
Fig. 3. The FC for explaining the changing point in the mar- Quinlan, J. R. 1986. Induction of Decision Trees. Mach. Learn. 1,
ket, analogically used for earthquakes. 1, 81–106
97