Preface
Computing and Information technology got part of our everyday life
within a generation. This is a dramatic change even more profound and
faster than the industrial revolution. Naturally it raises several
important questions which are being discussed and will be discussed in
several forums. I think the most important question is to answer how
computing can support the development and strength of our society and
how people can use it in the most effective way.
What differentiates a society from a group of individuals that members
of a society can afford specialization and is able to cooperate in a
very effective way. Naturally it needs rules and bounds to stream
energies into the right direction and protect the society against non
conforming individuals. However there is a single major benefit which
is makes worth all this effort to build and maintain a society; the
sharing of knowledge. A group of effectively cooperating people having
different skill sets and knowledge is more effective than any expert. (See „The Wisdom of Crowds” by James Surowiecki; http://www.youtube.com/watch?v=U0LhDQD7-ms)
It doesn’t mean that we don’t needs experts, scientists and gurus.
However what we need is an environment where those gurus can
concentrate on what they really know the best and let others to deal
with the „details” where they can excel. That is cooperation and sharing
best practice.
However IT is still often run by gurus. To make a new report, implement
an algorithm or put together a website we often rely on gurus. That is
bad because software is nothing else than packaged knowledge (See: „The
one minute risk assessment tool” by Amrit Tiwana and Makr Keil; http://www2.cis.gsu.edu/dmcdonald/MBA8120/Session5/RiskAssessmentTool_CACMFall2006.pdf) and if you can’t package and publish this knowledge you suffer a
disadvantage. Other disciplines have similar problems (e.g. to publish a
book you need a publisher) but the promise of internet is the easy
access and delivery of information. (See „The
long tails” by Chris Anderson; http://www.youtube.com/watch?v=LlAZ9t2m7-E&feature=related )
My objective with this paper is to explore to opportunities and tools for more effective knowledge share.
Methodology
The simplest methodology is used, I simple go through the traditional,
existing and potential methods of knowledge share and draw some
conclusions. In Internet I focus on text documents as a primary means of
knowledge share and see new methods as an extension of it. This is a
deliberate limitation, although other, very successful methods of
knowledge share exists (e.g. eLearning and Video) I think discussing
them won’t add much to the topic at hand. Maybe next time.
Knowledge Share
The Traditional Way
Since
prehistoric times sharing knowledge is the basis of civilization.
That’s not a surprise, people started on live in groups because so they
could share task, combine their skills and so be more effective.
However it’s not possible if they don’t share their knowledge. There
are two basic ways of sharing knowledge; the first by using the
knowledge for the benefit of others (for some – usually – material
direct or indirect reward), the second is to teach others to things you
know. In the latter case the reward is often more indirect.
The
thing we call job, or work belongs to the first category, Books,
teaching, practice, conferences are in the second category. Work groups
are somewhere in between.
Books
Historically
using books for information access got widespread only at the end of
XIX century. Before that it was only a privilege of a selected few. Up
to now books are the most used source of information. For a physical
storage media they have relative large capacity, mobility, are durable,
easy to handle and usually give a nice feel. On the other hand compared
to electronic media they space requirement is large, they are relative
hard to search and read only; it’s hard – albeit possible – to add
personalized content to them .
Teaching (http://www.adprima.com/teachmeth.htm)
Teaching
is a more tradition way of knowledge share, actually it exists since
the beginning of mankind, and parents always teach their children. Even
the most primitive culture uses teaching as a knowledge sharing tool.
Teaching is very effective (more effective than books) mostly because it
has immediate feedback, can be tailored to the people receiving
knowledge, and it’s easy to combine with other methods; books,
exercises…. However personalization is what limits the power of
teaching; with the number of participants its effectiveness is
declining, it needs infrastructure, and participants have to make an
effort to be at the same place in the same time. Even if it’s one to one
it requires constant alignment and concentration form pupil and teacher
(you can set a book aside if you e.g. feel tired, a lesson will be lost
if you can’t concentrate).
Practice
Learning
by doing is a thing we always do, however not very effective. However
learning by doing is very effective if a tutor can support the process.
Conferences
Conferences
are very similar to teaching with the difference that it’s usually
about new, not yet settled knowledge and the share of knowledge is more
two way. The presenter not only wants to share what she has found,
developed, concluded… but also is interested in the feedback of the
audience.
Work-groups
Work-groups and teams are very popular to solve complex tasks because members can
deal with the tasks with different points of view using their special
expertise and knowledge. However this efficiency comes at a price; there
is always an overhead even where the cooperation culture is strong (and
in some countries or workplaces it’s not the case), administration is
needed to manage the work, meeting has to be organized, task has to be
broken down and somebody with ample authority has to deal with the group
dynamics. A wrong team can be contra productive.
The Modern Way
eBooks and electronic publications
Electronic
books are the next generation books. Nowadays everybody – who has some
valuable topic to write about - can compose a document and publish it in
practically any format. However that doesn’t guarantee accessibility
(in contrast a publisher guarantees that it will at least distribute
your book in its shops), this I will discuss later.
The
other point is that a eBook could and should deliver more than a
“normal” book. It can be put in context i.e. reference to other
materials, text, videos, dictionaries, thesaurus, the user can add and
share his comments, notes and bookmarks.
In
other worlds content enrichment is key element in digital publications;
this is an important added value compared to traditional publications.
The following content enrichment methods and tools are feasible:
- References: links to other documents, publications, websites.
- Dictionary: in a narrow sense it’ allows interpreting, translating and exploring words found in the text. In a broader sense it allows to navigate through a thesaurus and explore a broader set of related documents.
- Search: a good solution would not only look up terms in the document itself but look for results also in the referenced documents.
- Similar documents: look for similar documents, web pages… etc.
- Text/Data mining: a clever algorithm could find keywords, concepts, assessments and data in the text. This would help to formalize intrinsic knowledge and also give good feedback to the author.
Distant learning
Distance
learning is – in may view – is very similar to traditional learning
only that it’s able to break over geographical and time barriers.
Personal contact is and stays the most powerful way of teaching but
it’s also very resource intensive (buildings, organizations, travel,
administrative overhead…). A good mixture of those can be the most
effective way of learning.
(Distant learning is very similar to telework.)
Software
Software
in its core is nothing else than knowledge; structure, work flow,
algorithms, rules… Plus content provided by the user(s). The history of
the last decades shows how powerful this concept is. However there is
one serious difficulty; the implementation of knowledge is tedious,
resource intensive and requires a programmers who must understand the
knowledge to be packaged. That‘s a major obstacle because sometimes even
experts have difficulties to formalize they knowledge not mentioning to
transfer it to professionals of another field.
What
we would like to have is something like an application blog e.g. where a
tax expert could not only write about the new tax calculation method
but also publish a tax calculator for the purpose. This question of user
programming is also important in traditional (like ERP) systems.
Web
It
doesn’t make much sense to praise the importance of the Internet.
Others have done it very profoundly and we all are aware of it. What is
important for us that while the Web opens the door for easy publications
and knowledge sharing for the masses it also raises questions
concerning the deluge of information, credibility and security.
Blog
Blogs
are a simple and easy way to share thoughts and follow events. However
by nature blog entries are short, rather like a piece of news. Blogs
can raise interesting thoughts, questions, direct interest to a topic
but can’t give deep knowledge.
Wiki
Wikis are probably the best way to collect knowledge in a certain area (see Wikipedia).
Website
A website is a complex thing, it can contain any kind of content or application.
The Twenty First Century Way
Semantic Web
The
semantic web (sometimes called Web 3.0; „The
semantic web” http://www.youtube.com/watch?v=OGg8A2zfWKg
) is an aspiration to give dumb
content a meaning. While a text on a web page has no meaning for the
computer the semantic web can classify text, sentences or keywords,
according to some classification. For example the sentence “Semantic web
is a new computing paradigm.” has no meaning for a computer. However if
you attach the classification “information technology” to it than
computer will have an easier job to attach search results, related
articles, user preferences to it.
The
difficulty of semantic web is the construction of the semantics
themselves. First it puts a “burden” to the publisher who has to enter
the semantics into the content. Second it needs a common understanding
and structure of semantics (i.e. metadata). I think that these
obstacles also give the limits of the semantic web. People publishing
on Web will give keywords, labels and some metadata to their texts (or
other content) but it’s not very probable that they will be able and
willing to use a complex thesaurus for every word or other pieces of
content their produce. Next I don’t think they will ever be a common
thesaurus for every domains of knowledge. Rather there will be several
different thesauruses for every domain it’s not very probable (and
neither desirable) to have one single “structure of knowledge”.
And
last the semantics will be produced by humans so it seems that we just
push the problems to a higher level. If people would phrase the
sentences more clearly no semantic web would be needed, search engines could
retrieve the semantics without major difficulty. (Although
the relative importance of text on the web is declining I discuss only text
publications here. However the some rules apply for pictures, videos and
applications in general.) And actually this is
happening now. (For a
good example see http://www.silobreaker.com/
)
To
summarize the semantic web indeed makes sense, it would help to
increase the value of content, improve knowledge share in certain
domains significantly but it won’t bring a fundamental change in the way
and efficiency we use the Internet.
Problem databases
Wikis
(especially Wikipedia) are an excellent and very popular knowledge
sharing tool. However it’s passive content only. What I would prefer is a
Wiki which is also support algorithms in a uniform way. Such a database
for example not only describes the traveling salesman problem but also
implements the algorithm in a standard manner.
Publication platforms
There
are plenty of web publication and content management platforms. However
there are limited for work-groups and companies. We need something more
open and common much like a social network.
Access to knowledge
The credibility problem
Credibility
is a long standing problem of knowledge sharing – just think of fake
degrees or plagiarism scandals. However couple of decades ago you could
trust the sources of information or knowledge to a certain degree.
Education institutions had to meet accreditation criteria, newspapers
employed well educated, experienced journalists with a track record and
professional books were reviewed by an editor. In general there was
enough stability and barriers of knowledge sharing that only
professionals could afford to share knowledge on a large scale and
credibility was easier to check and control. Naturally this world was
far from ideal, it was just much easier to find your way among the
sources of information. Now not only the amount of available information
exploded but so did the sources of it as well. New things arrive and
new sources of information appear and disappear. How do you know which
to trust? Is page rank
a good measure of credibility? Or are much "liked" pages credible? They
are a measurement at least, but not too reliable. From the
publishers/authors point of view credibility must be build, managed and
protected even more carefully on the Web than in life. In life you can
expect that time heels your early mistakes, on the Web these mistakes
may be only a click away. Character is even more important than before
even if that character is virtual. I don't think it's easy to maintain a
virtual character different from your own for a long time, not to
mention moral.
As
a reader it's even harder to decide which source to trust. Some sites
have a well earned reputation but several not. It would be good to build
a credibility index (where the creditors are measured as well) like digg, del.ico.us, StumbleUpon..
and so on. A credibility rank can be also imagined similar to page
rank. The credibility rank (CR) could depend on sources referred; a
higher number of credible sources referred would result in higher level
of credibility, and agreement; CR of a site is higher if other
(credible) sites dealing with a similar topic come to comparable
conclusions.
Naturally
every human and organization wants to build goof relationships but it’s
time to put credibility on your agenda, be credible also for people who
you even don’t know!
Shops
Subscriptions
If
you register to a website and pay for the access you have a
subscription exactly like to a newspaper. Newspapers are actually
information brokers, they collect, select, organize and distribute
content. There is a broad specialization among publications from general
magazines down to very specialized small circulation publications. As a
reader I can expect certain quality regarding the level and spectrum of
articles.
There
are two problems with this model; the first is (revenue) free access
via search engines and content aggregators and second the limited
personalization.
Free
access may make traditional journalism unsustainable, which is
problematic because it may lead the loss of the values of traditional
journalism. Although free journalism (e.g. blogs) often give better
information on certain topics and details but professional journalists
and editorial teams can give a better overall quality and coverage. (No
wonder that as time is going on the structure of web site teams are more
and more similar to traditional editorial organization). However free
information is threatening this model and we may lose something on the
way. If there is need for quality journalism in the future I expect
that content creation and content aggregation will come apart in the
future and done in separate businesses in the same supply channel.
The
limited personalization is the other problem. In a media business a
venture either can publish a journal with broad coverage or a series of
specialized journals. Both have its natural limitation; either we try to
reach everybody and not really satisfying anybody or go into every
niche which is not very economical bellow a certain level.
Web
sites try to resolve the limited personalization problem by giving the
user freedom to select from different sources and styles. This approach
also has its limitations:
- Customers may find difficult to find all relevant sources
- There is no system for source qualification (e.g. topic, subtopic, quality, reliability, length…)
- The solution is static; for the user it narrows the potential sources and type of content which will be a barrier for the user to broaden her knowledge and find new, interesting information.
I suggest a three level model:
- Content providers
Content
providers produce content (inclusive applications) probably in several
format (long study, professional article, poplar article, news) and
languages, and also take care for updates if needed.
Secondary
content providers may also be involved e.g. translators, experts doing
content enrichment, programmers implementing algorithms and models…
- Content aggregators
Content
aggregators qualify the incoming content, regarding topic, subtopic,
geography, importance, quality, reliability, importance, style, length,
difficulty…
- Customer support system
User
may access the content typically through a portal and give their
preference accordingly. However a recommender system makes a deeper
profiling, even considering the time (e.g. during the day, in the
evening, on the weekend, winter/summer… It also takes care to propose
new type of information and the opportunity for feedback.
eShops
Strictly
speaking electronic shops are only a way to sell you something.
Basically there are two way of shops; the large shopping mall and the
small grocery at the corner. Shopping mall became popular in the
seventies and eighties but now it seems that they are over their life
cycle. (See: http://www.economist.com/node/10278717?story_id=10278717
or http://deadmalls.com/)
Small shops lost a lot of their appeal in the recent decades and we
don’t know whether there will be a revival. However they still have the
advantage of personal touch, provided the personal is doing its job
well.
Web
shops don’t have physical limitations but they are even more impersonal
than a shopping mall. And exactly that is the point where improvement
can and should be done. The first step to make it happen is building
credibility. Users have to trust your site, the information and product
we sell. Next we have to offer quality. Quality means beside of 24x7
hours of operations, speed and hit rate also familiarity, that is the
feel of being home, finding our way easily among the zillions of
products, actions… etc. I think we should be careful with
personalization it can step by step narrow the choices A logical step
would be to mix small shops and web-shops to give personal experience
and the joy of shopping while keeping the huge selection and easy access
to web-shops.
Appstore
Technically
it’s not easy to understand the popularity of Apps. A well written web
app can give a similar user experience without all the hassle of
implementing, porting, installing, updating… a specialized application.
On the other hand apps have some benefits; they can use the local
resources better, providing a better user experience and also work off-line. However I think that the feel of ownership is at least as
important. (Unfortunately
I haven’t yet seen any proof or analysis about it.)
Community
Social networking is enormously successful. What I miss is a community site which supports real collaboration and can support - even ad-hoc - work-groups efficiently or crowdsource tasks. Social networks have the potential for it but they still have to deliver.
Search
Search
is a primary access interface to the web now. It’ success is due to the
huge amount of information which can’t be effectively managed with
other – more manual – methods. Search is automatic way to access
content and as such is very new. It’s widespread use drives innovation
in search and as a result search is partially replacing other access
methods;
- Content qualification: search engines are able do differentiation between more and less relevant content.
- Navigation and linking: search engines are able to collect similar documents and give references to them. It also means that search engines are able to build taxonomies and recognize keywords.
- Query/question: although search engines are not prepared to answer questions their query analysis features are good enough to interpret questions and return relevant content.
- Personalization; search is able to recognize personal preferences in search.
Search has huge advantages:
- Unlimited access; search engines can process any amount of information. (At a cost, but it can be done).
- Fast update; new search engines analyze and index new documents within hours after publication (or even faster).
However there are some drawbacks too:
- As indexes are made by machines their intelligence is limited
- Targeted search is often difficult: search engines try to give you more options and so always broaden the results even if you want’s to make it more specific.
I
think search will remain a dominant interface for information access
also in the future and stay at the frontier of development. What will
probably change is the search interface; search will be more often
“disguised” as navigation and linking or even mimic a natural language
interface
Navigation
Navigation
and linking is the classical web interface (web documents are hypertext
documents containing links to other documents), in the first version of
HTML linking was the preferred and only way to look up information.
However this is not very effective if you wants to look up information,
because it’s time consuming to go through all the links which not
inevitably leads to the information which you are looking for. So
portals emerged (which are nothing else that documents containing
organized links) to structure content and ease navigation.
Linking and navigation have the following benefits:
- Navigation gives an easy to understand and straightforward way to access relevant content
- Linking connects related information; it gives deepness to information, allowing staying on a higher level or digging deep as far as you want.
In
general linking and navigation supports information discovery,
navigation on the concept level, linking on the document level.
However linking and navigation have its limitations as well:
- It’s manual work which is time and resource consuming.
- Quality is mutable
- Targeted access – when you are looking for a certain piece of information - is tedious
Query
Research trail
In
the time of the big discoveries explorers had to beat through unknown
territories and be prepared for the unexpected, face dangers, overcome
obstacles, fail and restart – all this with an unbelievable effort.
People who followed them had a much easier task, they could follow
their “trail” of their predecessors.
Research
trail is the path a researcher follows during her journey through
information, documents, tasks, ideas, experiments… Recording the trail
of the research is very useful to document how one came to a conclusion
and what steps had he done. This is a very common thing, almost
everybody doing research take notes, logs…
If
we do research on the Internet (in practice all of us do) we
navigate/search from site to site. Technically it’s no difficulty to log
the trail of our research. Let’s assume that we can share this trail
with others and there is a database of such logs. Let’s also assume
that if you enter a search expression or open a website a tool shows you
in which direction other people entering the same search expression or
opening the same website “went” from there. You could see the major
routes and also where they lead. E.g. by entering the word “Chickago”
into the searchbox the most popular, travel, hotel, culture, history,
political sites would be displayed to me. (It’s also called social search. This is a
good example, albeit the site seems to be dead: http://blog.researchtrail.com/ )
Personalization
Personalization
helps people to define their particular needs and interests and have
more effective access to knowledge. However it also may confine the user
to the sources of information he used before and this can be contra
productive because 1. People like (and need) to discover new things.
Sources and people change and personalization may damp this process.
Recommender
Recommender
is in essence automated personalization. It requires no extra effort
from the user and is more flexible. On the other hand its use is limited
to showing information which may be of interest for the user. With
time there will be diminishing difference between personalization and
recommender.
Web database
Database provide effective access to structured data.
New business models
Advertisement
Now
this is the business model of the web for knowledge and information
share. Google and other large web portals use this model. This is not a
new thing, broadcasting (TV, radio) and newspapers also rely on
advertisement as the primary source of income. Web has the advantage
that it can align adds to context and user preferences providing an even
more targeted adverts.
Although a very straightforward and clear model, adverts also have their drawbacks:
- Ads take away time and place from real content
- Adds income is – often – realized for search companies and portals and not for the primary content producer
As
we see the first problem is the concern of the customer. For them the
solution could be a subscription service. The second hits the content
providers; in a similar manner they also can ask a fee for “reusing”
their content.
Donation
That’s
how Wikipedia works and this provides the best user experience (the
only ads are the ones where the site asks for donations). In my view
the only limitation of this model is that people – who earn their money
in a different way – spend money only on projects which they think serve
some public good. (And they have fairly right of doing so!).
Web shop
Appstore
Apple
made this kind of distribution popular and now everybody who has an
acceptable market presence is building something similar. The benefits
seem to be obvious; developers have an excellent distribution channel,
there is a certain level of quality and security maintenance, the
applications is tailored to the platform and customers have a broad
selections and the feel of ownership. Modern applications can cleverly
share local and cloud resources too.
Technology
Databases
Databases
give strong structure to data which opens the door for accessing and
processing large amount of information and transform raw data to
information (e.g. through data mining). The price – no wonder – is the
lower flexibility and increased investment into knowledge. If data
embedded in text could be made explicit, the access to knowledge could
get a boast. There are several ways to do it, here some examples:
Data recognition and exposition
With
refined linguistic knowledge the data (places, companies, numbers,
qualifiers… etc.) can be recognized in text. My favorite example is
Silobreaker (www.silobreaker.com)
which is able to categorize data found in text. In fact Silobreaker is
rather about classification and connecting related information than
date exposition, but it show how hidden information can be made visible
to a broad audience. However even this wonderful solution is can’t make
all data visible and sometimes makes mistakes. The solution could be to
make every data explicit by creation.
Semantic Web
Simple
speaking the objective of the semantic web is to give meaning (i.e.
expose) the data embedded in the text. Using that data new information
sets can be generated, complex searches are possible among different web
pages and domain and automatic agents can look up complex information.
Semantic
Web is not a new concept and considerable effort has been put into it.
Still – in spite of some good examples – I can’t see many semantic web
applications. It doesn’t mean that the semantic web doesn’t make sense
or that there is no progress. I just say that the progress is slow and
maybe we miss something. What it can be I will discuss later. (See:
Wikidata
Wikidata
is a good example of the semantic web in practice. We also can see that
a strong and well organized community is setting up relative modest
targets. In the first step the objectives is replacing data found in
texts with links to a database and share it among entries. The practical
benefit is that translations can use the same database. The future is
not yet clear – there is not such a rigorous target setting as in case
of the semantic web, however once the database will be there it can be
used as source of automatic processing and further information services.
The
Wikidata project is very interesting because it set very realistic
targets i.e. it will be successful and second during implementation it
will face exactly the same problems which prohibit the fast spreading of
the semantic web. It’s a real word experiment.
Web Database
The
semantic web and Wikidata both start from text. However the reverse
approach is also possible i.e. building a database of a domain (e.g.
company information, hotels database or legal database). Several of such
databases exist and used everywhere. This approach naturally also has
the classic limitations; they are usually confined to a geography,
language, domain of knowledge and modeling approach. They are also
often limit access to subscription.
http://meta.wikimedia.org/wiki/Wikidata
http://techcrunch.com/2012/03/30/wikipedias-next-big-thing-wikidata-a-machine-readable-user-editable-database-funded-by-google-paul-allen-and-others/
Software tools
People
need new kinds of software tools to publish their knowledge in more
structured form than text, pictures or video. Two major direction seems
to be feasible, the first tries to make programming as easy as
possible, the second tries to implement tools which are near to the
tools professional – who are not programmers – use.
Encapsulation
The
major advantage of software is that it can encapsulate knowledge for
the user. While text (picture and video) can very good of transferring
knowledge (although software also can use very well for the same
purpose) software makes knowledge usable even if the user don’t know
the details of the solution, while text requires you to at least go
through an algorithm. (The difference is like receiving all the parts
and full documentation to assemble a car or receiving the car ready
made). The ideal solution would be to mix code and text – which can be
done e.g. with HTML but not naturally.
Scripting languages
Scripting languages# are
very much like traditional languages; de major technical difference is
that scripting languages are interpreted and not compiled. However it’s
not easy to give an exact definition of scripting languages; in general
they try to make small programming tasks easier and faster and often
are made to support a specific task (e.g. shell programming) or be a
“glue” language to connect different tools, resources and “heavy”
languages.
There are a large selection of scripting languages, notable examples are Python#, bshell#, JavaScipt#, Perl# and Tcl#.
http://www.mactech.com/articles/mactech/Vol.15/15.09/ScriptingLanguages/index.html
http://www.perl.org/
http://www.mactech.com/articles/mactech/Vol.15/15.09/ScriptingLanguages/index.html
http://www.perl.org/
Scripting
languages are better suitable for user programming than “real”
programming languages but they still programming languages. They provide
a higher level of abstraction and are easier to use but the difference
to programming is not significant.
.
perl 
python 
javasript 
tcl
perl python javasript tcl
Google trends show a slowly diminishing interest for scripting languages.
Functional languages
Functional languages#
differ from “traditional” programming languages that they describe
functions rather than imperative instructions. That makes them very
effective especially in expressing recursive operations. Well known
functional languages are Common Lisp, Scheme , ISLISP, Clojure, Racket, Erlang, OCaml, Haskell, Scala and F#.
[1] http://en.wikipedia.org/wiki/F_Sharp_%28programming_language%29
[1] http://en.wikipedia.org/wiki/F_Sharp_%28programming_language%29
Functional
programs are very useful in certain areas and for educated programmers
of professionals, but they are not very suitable for people who are not
programmers.
Functional programming
Domain
specific languages are by nature very powerful in the domain they are
written for. This may be very tempting for specialists of the field to
invest into the learning of the language.
Macros
This
is a simple thing, but I wish that equations could be exposed to a
solver. E.g. if I write the tag <EQ>y=sqr(x); y:”square”,x:input”
<\EQ> it will be exposed to a simple input field called “input”
and a result field called “square”, the style defined in stylesheet.
Decision table
Decision tables#
are the next step from data to algorithm. The advantage of decision
tables is that they are easy to construct (-provided that the
appropriate knowledge is available) and easy to use. Decision tables can
also be nested or extended with other knowledge representing tools.
However it’s not easy to “grow” a decision table above a certain
complexity.
#
#
Decision trees
Decision trees# are similar to decision tables, the major difference is the graphical representation.
Genetic algorithms
I
don’t feel confident enough to analyze the possibilities of genetic
algorithms and similar tools (e.g. neural networks, petri nets,
simulated annealing, simplex, …). The important thing is that it is
possible to make tools which – if provided with appropriate input
information are able to “find” solutions to complex problems.
Agents
Agents
are software standalone applications which can solve different problems
on behalf of the user (which also means they “know” the user). They
collect information which is important for you, or do task on your
behalf. To be effective they need a large amount of semantic data or a
lot of artificial intelligence. Although the notion of Agents (they are
several types of software agents) is around quite a while I don’t see
many examples of personal agents around. (Simple agents like newsreaders
do exist).
software agent
#
software agent#
Personal computing
Personal computing is changing significantly (just think on the success of tablets). There are two directions of development.
- Personal computers start to dissolve and taking new forms. I think on devices like game controllers, smart phones, television sets. These devices will take over several functions from today’s computers but in essence they will be consumer devices.
- Those who produce content will need tactile input also in the future and full computing capabilities (word processing, calculation, programming, video edition…). These computer will also undergo a very intensive development however they will resemble stronger to laptops as we know them. They just will be lighter, flatter; will have much longer battery life, easier use and mobility, flexible connection to cloud services.
Server computing
Servers will move to cloud and data centers will disappear.
Performance requirements
For
“content producing” computers more flexibility is expected. While they
will sacrifice performance for long battery life when they have no
connection to power and to the internet, performance will boasted by
cloud services and local resources will be used intensive when the
computer is plugged. Computers will learn the users usage profile and
adapt to it.
Cloud
The
challenge of the Cloud to integrate seamlessly with personal computers
that the user can use cloud services and locate resource hungry
operations to the cloud while be able to run effectively when offline
(although it will happen not very frequently in the future).
Assessment
Which
is the best method? As with everything in life there doesn’t exists the
optimal knowledge sharing method. However we can set up a set of
criteria which can help us to select the optimal way and improve tools.
Criteria
- Credibilty#
Naturally credibility depends on the author but user must be in the position to decide how credible a source is.
- Creatibility
How easy is to publish (i.e. create knowledge).
- Accessibility
How easy is to access knowledge.
- Motivation
There must be a system which motivates authors to share they knowledge.
- Comprehensibility
Measurement
Conclusion
Process
The online knowledge production process looks like this.
If
we analyze the process as we would a traditional production process
then we find an imbalance. Almost all new initiatives like the semantic
web or new search technologies come from the distributor’s side and
most investment comes from there. However the new initiatives put the
burden on the knowledge production (lets call them publishers). They
have to define microdata, semantic information, links, taxonomies, etc.
We also would like that they fill in databases, write code, construct
tables, macros…
I
draw two consequences from this facts; the first is that publishers get
better and more powerful tools, the second that publisher should have
stronger influence on the technology and content ownership.
Problem structure
There are the following problems with online knowledge share.
Cooperation
There
are a growing number of websites and that is a good thing. The failure
of socialism has shown that it’s not possible to have a single, super
rationalistic structure even for a single domain of knowledge, industry
or market.
However
from a consumer perspective an easy navigation is needed between
websites, applications , documents… That requires that these resources
of knowledge speak “the same language” they can be interlinked. This is
the promise of the semantic web, it connects information on the meta
level. However it’s not possible to have a single set of meta-data, there
always will be an area not covered or a perspective not regarded. The
only thing we can expect that these meta-data will have some common touch
points very much like the HTML documents can link to each other.
Proposal
In my view the ultimate publication process will look like this.
- It starts with the consumer; authors, editors and alike will permanently check their consumers for their information needs. Naturally it’s done today as well, but I expect stronger tools than before.
- The basis for publication is text as today.
- A refined authoring tool defines the domain of the document (if the author hasn’t), looks for similar documents, references, connections… This feature is build into the text editor; the software runs in the background and runs permanently.
- Software chooses an appropriate taxonomy and identifies entities in the text. User can modify these selections (and also extend the local taxonomy).
- The tool identifies places, data and maps it into a database. Again the user can modify data or enter new data. The software can only work with predefined database schema.
- Publisher also can use and define software tools in her text (macros, decision tables…). These are also stored in a database.
The question is who pays for the content. Again I have the problems that revenue is moving away from authors which may lead for a decrease of content quality and in the amount of good content. Now publishers can either choose an Appstore model, rely on advertising or subscriptions. In may point there should be a stronger specialization where publisher produce the content and distributors pay for it (much like an average grocery store cooperates with it partners). This is a well proven model of the non digital world.
No comments:
Post a Comment