Authorea

Joe Corneli tone down our 'recommendations' about 9 years ago

Commit id: 519ae03bf81b962832e7e05118413a60c31dd898

deletions | additions

@phdthesis{eric-nichols-thesis, title={{M}usicat: {A} {C}omputer {M}odel of {M}usical {L}istening and {A}nalogy-{M}aking}, author={Eric Paul Nichols}, year={2012}, school={Indiana University}} @phdthesis{kaz-thesis, title={Interpretation-driven {A}ssociation in {D}esign}, author={Kazjon Grace}, year={2011}, school={University of Sydney}} @article{helms2012analogical, title={Analogical problem evolution in biologically inspired design}, author={Helms, M and Goel, A}, journal={Design Computing and Cognition, June}, volume={9}, year={2012} } @incollection{Analogical-problem-evolution-DCC, year={2014}, isbn={978-94-017-9111-3}, booktitle={Design Computing and Cognition '12}, editor={Gero, John S.}, doi={10.1007/978-94-017-9112-0_1}, title={Analogical Problem Evolution in Biologically Inspired Design}, url={http://dx.doi.org/10.1007/978-94-017-9112-0_1}, publisher={Springer Netherlands}, author={Helms, Michael E. and Goel, Ashok K.}, pages={3-19}, language={English} } @article{sitkin2011paradox, title={The paradox of stretch goals: Organizations in pursuit of the seemingly impossible}, author={Sitkin, Sim B and See, Kelly E and Miller, C Chet and Lawless, Michael W and Carton, Andrew M},

Year = {1951}} @book{bergson2010creative, Annote = {original title: La {P}ens{\'e}e et le {M}ouvant}, {M}ouvant, Trans. Mabel L. Andison.}, Author = {Bergson, Henri}, annote = {Trans. Mabel L. Andison.}, Place = {Westport, Connecticut}, Publisher = {Greenwood Press}, Title = {{T}he {C}reative {M}ind},

year = {2010} } @unpublished{Adamopoulos2011, author = {Adamopoulos, Panagiotis and Tuzhilin, Alexander}, file = {:Users/worldwindow/Documents/Mendeley Desktop/Adamopoulos, Tuzhilin - 2013 - On Unexpectedness @article{Adamopoulos2011, title={On unexpectedness in Recommender Systems recommender systems: Or How how to Better Expect better expect the Unexpected.pdf:pdf}, mendeley-groups = {Magisterarbeit/Recommendation unexpected}, author={Adamopoulos, Panagiotis and Serendipity}, pages = {1--50}, title = {{On Unexpectedness in Recommender Tuzhilin, Alexander}, journal={ACM Transactions on Intelligent Systems : Or How to Better Expect the Unexpected}}, year = {2013} and Technology (TIST)}, volume={5}, number={4}, pages={54}, year={2014}, publisher={ACM} } @inproceedings{Kamahara2005,

year = {2005} } @inproceedings{Onuma2009, author = {Onuma, Kensuke and Tong, H and Faloutsos, Christos}, booktitle = {Proceedings of the 15th ACM SIGKDD international conference on Knowledge discovery and data mining}, file = {:Users/worldwindow/Documents/Mendeley Desktop/Onuma, Tong, Faloutsos - 2009 - TANGENT A Novel 'Surprise-me' Recommendation Algorithm.pdf:pdf}, isbn = {9781605584959}, mendeley-groups = {Magisterarbeit/Recommendation and Serendipity}, title = {{TANGENT : A Novel 'Surprise-me' Recommendation Algorithm}}, url = {http://dl.acm.org/citation.cfm?id=1557093}, year = {2009} } @inproceedings{Oku2011, author = {Oku, Kenta and Hattori, F}, booktitle = {Proceedings of the Workshop on Novelty and Diversity in Recommender Systems (DiveRS 2011), at the 5th ACM International Conference on Recommender Systems},

\subsubsection*{Key condition for serendipity} Serendipity relies on a reassessment or reevaluation -- a \emph{focus shift} in which something that was previously uninteresting, of neutral neutral, or even negative value, becomes interesting. \begin{itemize} \item \textbf{Focus shift}: ``\emph{After removing several of the

\subsubsection*{Components of serendipity} A focus shift is brought about by the meeting of a \emph{serendipity trigger} and a \emph{prepared mind}, which is then involved in mind}. The next step involves building a \emph{bridge} to a valuable \emph{result}. \begin{itemize} \item \textbf{Prepared mind}:

\subsubsection*{Dimensions of serendipity} The four components described above have attributes which that may be present to a greater or lesser degree. These are \emph{chance} are: \emph{Chance} -- how likely was the trigger to appear?; \emph{curiosity} \emph{Curiosity} -- how likely was this trigger to be identified as interesting?; -- \emph{sagacity} \emph{Sagacity} how likely was it that the interesting trigger would be turned into a result?; -- and \emph{value} \emph{Value} (how valuable is the result that is ultimately produced?). \begin{itemize} \item \textbf{Chance}: Fleming \citeyear{fleming} noted: ``There are

\subsubsection*{Environmental factors} Finally, serendipity is made seems to be more likely for agents who experience and participate in a \emph{dynamic world}, who are active in \emph{multiple contexts}, occupied with \emph{multiple tasks}, and who avail themselves of \emph{multiple influences}. \begin{itemize} \item \textbf{Dynamic world}: Information about the world develops over time, and is not presented as a complete, consistent whole. In particular, value \emph{value} may come later. Van Andel \citeyear[p. 643]{van1994anatomy} estimates that in twenty percent of innovations ``something was discovered before there was a demand for it.''

Here we do not mean to suggest that every instance of ``a solution to a problem in a context'' is due to serendipity at work -- on the contrary, that is just the discovery step. Inventing a viable design pattern only happens when the solution is found to be explicable and useful. To van Andel's assertion that ``The very moment I can plan or programme `serendipity' it cannot be called serendipity anymore,'' we

online system that gathers new modules over time, and for which, periodically, new combinations of modules can yield new and interesting results -- already described in outline form in \cite{pease2013discussion}. After % Developing experiments along these lines may help to develop the groundwork for the more theoretically-involved excursion of involved designs discussed in the current paper, returning to simple experiments with this sort paper. % Patterns of system is our next planned step. Representing, and learning, design patterns in programmatic terms is a ``stretch goal'' -- but patterns serendipity, like the one one, above will offer useful heuristic guidelines for human programmers right away, working towards this goal, and give an idea also helps to convey a sense of our long-term plans. plans for fully automated discovery+invention. % Is ``having a stretch goal'' an example of a serendipity pattern? I think so!

\subsection{Future Work} \label{sec:futurework} \label{sec:hatching} As a framework for managing the twinned processes of discovery and invention, we are drawn to the approach taken by the \emph{design pattern} community \cite{alexander1999origins}, although \cite{alexander1999origins}. The essential features of this approach are described below, but we should point out straight away that we propose to use design patterns in rather nonstandard way: manner: \begin{itemize} \item[(1)] We want to encode our design patterns directly in runnable programs, not just give them to programmers as heuristic guidance.

not just capture the solutions to existing ones. \end{itemize} \citeA{meszaros1998pattern} describe the typical scenario for authors of design pattern writers: patterns: ``You are an experienced practitioner in your field. You have noticed that you keep using a certain solution to a commonly occurring problem. You would like to share your experience with others.'' They There are many ways to describe a solution. Meszaros and Doble remark, ``What sets patterns apart is their ability to explain the rationale for using the solution (the `why') in addition describing the solution (the `how').'' Regarding the criteria that pattern writers seek to address: ``The most appropriate solution to a problem in a context is the one that best resolves the highest priority forces as determined by the particular context.'' %% Their article describes a number of criteria relevant to writing %% good design patterns, e.g. \emph{Clear target audience}, %% \emph{Visible forces}, and \emph{Relationship to other patterns}. A good design pattern \emph{describes} the resolution of forces in the target domain;at least in the setting we're interested in, creating a new design pattern also \emph{effects} a resolution of forces directly. The use case of design pattern development maps into our diagram of the basic features of serendipity as follows:

% An interdisciplinary perspective on the phenomenon of serendipity promises further illumination. Here, we consider the potential for formalising this concept. This paper follows and expands \citeA{pease2013discussion},which is a robust brief survey where many of the ideas that are developed here were first presented. The current paper uses the opportunity of working with a fresh start and a wider broader canvas to advance some bold claims about the usefulness of serendipity as anew framework for computational creativity. Serendipityitself is itself centred on reassessment. For example, a non-sticky ``superglue'' that no one was quite sure how to use turned out to be just the right ingredient for 3M's Post-it\texttrademark\ notes. %

serendipity, and examine prior applications of the concept of serendipity in a computing context. Then in Section \ref{sec:background} we present our formal definition of serendipity, drawing connections with historical examples and presenting standards for evaluation. Section \ref{sec:computational-serendipity}then presents computational case studies and thought experiments in terms of this model. Section \ref{sec:discussion} offers recommendations for researchers working in computational creativity (a key research area concerned with the computational modelling of serendipity), and describes our own plans for future

\subsection{Recommendations} \label{sec:recommendations} Deleuze wrote: ``True freedom lies in the power to decide, to constitute problems themselves'' \cite[p. 15]{deleuze1991bergsonism}; von %% Von Foerster \citeyear[p. 286]{von2003cybernetics} advocated a \emph{second-order cybernetics} in which ``the observer who enters the system shall be allowed to stipulate his own purpose.'' Deleuze \citeyear[p. 26]{deleuze1994difference} wrote: %% \begin{quote} %% ``\emph{We learn nothing from those who say: `Do as I do'. Our only %% teachers are those who tell us to `do with me', and are able to %% emit signs to be developed in heterogeneity rather than propose %% gestures for us to %% reproduce.}''~\cite[p. 26]{deleuze1994difference} %% \end{quote} %% These perspectives present a number of challenges to the typical %% programming paradigm, which is linked far more closely to a %% \emph{first-order cybernetics} that specifies system goals and %% operations at the outset, and provides external evaluation. % Dewey, Whitehead similar too. Our thought experiment in Section \ref{sec:ww} explores these ideas, and helps to illustrate develops a design illustrating the relationship between problem creation creativity at the level of artefacts (e.g. new poems) and serendipity. creativity at the level of \emph{problem specification}. The search for connections that make raw data into ``strategic data'' ispart of this common ground, and an appropriate theme for researchers in computational creativity to grapple with. %% Problem %% evolution (by analogy to existing design solutions) is discussed in %% \cite{Analogical-problem-evolution-DCC}, focusing on the case of human %% designers. As \cite[p. 69]{pease2013discussion} remark, anomaly %% detection and outlier analysis are part of the standard machine %% learning toolkit, but it seems In \cite{stakeholder-groups-bookchapter}, we outlined a general programme for computational creativity, and examined perceptions of

-- understood as human communities. We should now add a fourth important ``stakeholder'' group in computational creativity research: computer systems themselves. Creativity may look very different to this fourth stakeholder group than it looks to us. We should give help computersthe tools effectively evaluate their own results andtheir creative process. % %% These ideas set a relatively high bar, if only because computational %% creativity has often been focused on generative rather than reflective %% acts. As Campbell \citeyear{campbell} writes: ``serendipity presupposes a smart mind.'' We may be aided in our pursuit by recalling Turing's proposal that computers should ``be able to converse with each other to sharpen their wits''

\subsection{Related work} \label{sec:related} An active research community investigating computational models of serendipity can be found in information retrieval \cite{Toms2000} retrieval, andmore specifically, in recommender systems. systems \cite{Toms2000}. In the latter this domain, Herlocker et al. \cite{Herlocker2004} \citeA{Herlocker2004} and especially McNee et al. \cite{McNee2006} promoted \citeA{McNee2006} view serendipity as an important factor for user satisfaction, next to accuracy and diversity. There are several definitions of serendipity Serendipity in this domain \cite{Lu2012}, which commonly recommendations variously require the system to recommend deliver an unexpected and useful \cite{Lu2012}, interesting \cite{Herlocker2004}, attractive or relevant \cite{Ge2010} item. In terms of the prior definitions, the problem can be framed as item \cite{Ge2010}. %% Recommendations are typically meant to help address the user's difficulty in finding items that meet his or her interests or demands within a large and potentially unobservable search space. This problem The end user can also be passive, and items are suggested to support other stakeholder's goals, e.g. to increase sells. Definitions differ in as to the requirement of novelty; \citeA{Adamopoulos2011}, for novelty, and some researchers \cite{Adamopoulos2011} develop example, describe systems for suggesting that suggest items that might may already be known, but are still unexpected in the current context. In terms of our model, these systems focus mainly on producing a serendipity trigger, but they include aspects of user modeling which could bring other elements into play. Paul Andr{\'e} et al.~\citeyear{andre2009discovery} look at have examined serendipity from a design perspective. These authors also propose a two-part model, in which what we have called \emph{discovery} above exposes the unexpected, while \emph{invention} is the responsibility another subsystem that finds applications. According to Andr\'e et al., the first phase is the one that has most frequently been automated, automated -- but they suggest that computational systems should be developed that support both aspects. Their specific suggestions focus on representational features: \emph{domain expertise} and a \emph{common language model}. Although tremendously useful when they are available, these features are not always enough to account for serendipitious events. Using the terminology we introduced above, earlier, these features seem to exemplify aspects of the \emph{prepared mind}. However, as we mentioned above, the \emph{bridge} is a distinct process that mental preparation can support, but not always fully determine. For example, participants in

outcome. They particularly focus on the acts of \emph{reflection} that foment both the creation of a bridge and estimates of the potential value of the result. % Although this description touches on all of the features of our model, {\sf SerenA} nevertheless largely matches the description offered by Andr{\'e} et al.~\citeyear{andre2009discovery} of discovery-focused systems: systems, and a case study of a recommender system focused on serendipity. Here, the user is the primary agent with a prepared mind. Accordingly it is the user that undergoes an ``aha'' moment and takes the creative steps to realise the result; the computer is mainly used to facilitate this. The {\sf SerenA}'s primary computational method is to search outside of the normal search parameters in order to engineer potentially serendipitous (or at least pseudo-serendipitous) encounters. %% Another %% earlier related example of this sort of system is {\sf Max}, created %% by Figueiredo and Campos \citeyear{Campos2002}. The user emailed {\sf %% Max} with a an existing list of interests and {\sf Max} would find return a webpage %% web page that may might also be of interest to the user. Similar interest. Other systems with similar %% support for serendipitous discovery involve searching for analogies %% \cite{Donoghue2002,Donoghue2012}) and as well as content \cite{Iaquinta2008}. In earlier recent joint work \cite{colton-assessingprogress}, we presented a diagrammatic formalism for evaluating progress in computational creativity. It is useful to ask what serendipity would add to this formalism, and how the result compares with other attempts to

progress with \emph{systems} rather than with \emph{problems}. It would be a useful generalisation of the formalism -- and not just a simple relabelling -- for it to be able to tackle problems as well. However, it is important to notice that progress with problems does not always mean transforming a problem that cannot be solved into one that can. Progress may also apply to growth in the ability to \emph{posit} problems. In keeping track of progress, it would be useful for system designers to record (or get their systems to record) what problem a given system solves, and the degree to which the computer was responsible for coming up with this problem. As \cite[p. 69]{pease2013discussion} remark, anomaly detection and outlier analysis are part of the standard machine learning toolkit -- but recognising \emph{new} patterns and defining \emph{new} problems is more ambitious. Complex analogies between evolving problems and solutions form one of the key strategies for teams of human designers \cite{Analogical-problem-evolution-DCC}. Kazjon Grace \citeyear{kaz-thesis} presents a computational model of the creation of new concepts and interpretations, but this work did include the ability to create new higher order relationships necessary for complex analogies. New patterns and higher-order analogies were considered in Hofstadter and Mitchell's {\sf Copycat} and the subsequent {\sf Metacat}, but these systems operated in a simple and fairly abstract ``microdomain'' \cite{hofstadter1994copycat,DBLP:journals/jetai/Marshall06}. %% More %% recently, similar architectures have been used in music %% \cite{eric-nichols-thesis}. The relationship between serendipity and novel problems receives considerable attention here, in the current work, since we want to increasingly turn over responsibility for creating and maintaining a prepared mind to the machine.

in other words, what they find in the presented work. In some settings this is augmented with {\tt suggestions}. After any {\tt questions} from the author, the commentators may make {\tt replies} to offer clarification.\footnote{We clarification. %\footnote{We return to discuss further work with Writers Workshops and serendipity in Section \ref{sec:futurework}.} This is how these steps map into the diagram we introduced in Section \ref{sec:background}: \input{ww-schematic-tikz}