Tagcrumbs vs. review sites

While I understand that tagcrumbs, being a startup company, attempts to clarify in what sense it is different from review sites such as Qype or Yelp, I still find that from a long-term perspective, there are more similarities than differences.

The core idea behind tagcrumbs, as I understand it, is that it enables you to associate any resource to a geocoordinate (also sometimes called a place). Whether you have been to a specific place does not matter, so it is not only about sharing experiences, but also meant as a kind of geotagged moleskine.

On the other hand, review sites such as Qype started out as a kind of yellow pages enhanced with user-generated content that is associated to local places. Most of the places are imported from yellow pages data to make it easier to write reviews, but there are little constraints regarding the generation of new places (even virtual ones, which I personally do not care much about, but it does not really matter here). Important additions were events (associated with time or time periods) and guides (i. e. lists of places which enables users to generate tours).

In the three years that Qype has been online, it is possible to say that the focus was expanded from pure reviews to experience sharing. To claim that

Tagcrumbs has no thematic focus on reviews and thus supports a higher diversity of user-generated content. It's about all the little local discoveries and the insider knowledge around us

is correct from Tagcrumb's perspective, but I do not see in what sense the diversity of user-generated content is higher. Does this refer to content types, or to the semantics of content? Regarding the former, I see a wide variety (except perhaps the association of external links to geocoordinates, which should be constrained anyway for security reasons), and regarding the latter, what you write about should have some connection to the place you are considering, but again, I see little constraints.

And even if the diversity of user-generated content was higher, what would be the advantage? I do argue that sites operating on the basis of user-generated content need to invest some editorial work to make sure that what is contributed is in accordance with some set of requirements stated by the owner of the site, and in addition provide sophisticated means of search and retrieval (fulltext, categories, tags, associations etc.). The more content you have the harder it will be to find what you're looking for.

To sum up, I would see Tagcrumbs and Qype as quite similar in concept and focus, with the difference that the former starts out large (allowing a greater variety in the first place) and, as time goes by, will see the need to narrow down its focus, while the latter started quite small and is expanding its focus in order to perhaps attract more users.

The Success of Crowdsourcing

Interestingly enough, the term crowdsourcing was invented by a journalist and not someone from IT or business. Of course, a must-read in that context (and a great overview, too) is Jeff Howe's Wired article form 2006 entitled The Rise of Crowdsourcing.

While outsourcing means delegating a task to a defined workforce outside one's own company, the workforce to deal with a crowdsourced task is not defined a priori - it could in theory be anyone connected to the Internet. Most delegated tasks involve either the creativity or the intelligence of the masses - where this, of course, does not mean the average factor of a given population, but the selection of those who can best fulfill the task, but with a substantially reduced budget. For companies involved in crowdsourcing, it's all about saving money and milking the masses for inspiration, but for non-profit ventures, it's about participating to get some benefit in return - perhaps not immediately, but in the long run. That's why Wikipedia works so well, but WikiaSearch doesn't.

All in all, three factors play a major role in crowdsourcing: money, reputation and fun. But this in itself is no guarantee that a crowdsourcing activity will actually work. Most importantly, define the task you would like to delegate and formulate it in the form of an open call, then decide who should be the participants in it (contributors and / or voters) as well as the success factors so you will be able to evaluate your success (read more about designing and evaluating crowdsourcing).

To sum up, crowdsourcing is, first of all, a buzzword. But applied wisely, relative to your business and after a careful design, you may very well benefit from it if you succeed to gain the participation of your potential (or actual) customers. Good luck!

Microblogging is not micromessaging

While I understand that from a technical perspective, microblogging is in fact micromessaging, I would rather argue that from a conceptual and communication-oriented perspective, the term microblogging for services such as twitter or jaiku is indeed more appropriate. The explanation may seem rather simple:

• In messaging applications, there is a sender defining a number of addresses. Typically, these applications are working asynchronously. For synchronous applications, the term to use would be conferencing, in analogy to meetings.
  
• In blogging applications, the originator of a message or contribution does not know about who will read, as there are no reading access restrictions. This also means that anyone may comment (or reply) to a contribution. With regards to weblogs, commenting may be moderated (i. e. the owner needs to approve a comment), or turned off (e. g. if discussions following a contribution should be suppressed).
  

That is, in a nutshell, why microblogging is in fact just that, and not micromessaging.

E-Books and their readers

I have often asked myself what the benefit of e-books and their associated hardware is. With incompatible formats (e. g. Amazon's custom format vs. Epub and Adobe's EBook format), being dependent on energy sources, I am convinced that there must be additional benefits. OK, an incorporated search function is a good point to quickly find what you are looking for, but there must be more. For instance, a personal profile that augments itself with every downloaded item and that would allow for personal recommendations of items to buy / download. Once newspapers will start offering their articles in e-book compatible formats, personalization and consideration of current context is another idea, even if the provider does not offer such a feature. Providing a link to social library services, such as LibraryThing, being able to not only read reviews, but also share book recommendations (with alerts from one's own social networks) seems to be interesting. The reader itself does not need to be pimped up a lot for that, it could also use a central broker service offering these features and needs to be able to provide appropriate input and output channels. With WLAN this does not seem to be a problem. So, what do YOU think is the future of e-books and their readers?

Future of News?

Newsmakers are increasingly going online, which started with selected articles from their printed publications, sometimes abbreviated versions to give an overview and offering full articles for an additional fee. In most cases, it is possible to leave comments, and sometimes, text is enriched with (inline) video contributions, which are sometimes excerpts from TV programs related to a given topic. In order to inform their clients, daily newsletters are being compiled and sent to interested users. Many newsmakers also participate in social networks in order to make their "brand" more visible so it is possible to become a "fan", whatever this means. Another useful channel to inform users about new issues is microblogging, where headlines are posted to e. g. Twitter with a link to a related article.

All in all (besides being able to comment), news is still a one-way business where contributions (text, audio, video) are being published, sometimes aggregated by topic. It is neither personalized, nor is there a link between paper and online news.

Enter 2D barcodes as a potential link between articles or ads and related resources. With regards to ads, Google has been experimenting with barcode-enhanced ads that lead to a result page aggregating results related to the product or resource being advertised, including e. g. some additional information on the related company. With barcode-enhanced printed articles, URLs to multimedia resources can be coded, along with other static information. Given a thin client on a mobile device, which is capable of gathering context information (e. g. time, location, weather) and able to access personal information (e. g. a user profile), the prerequisites are there to create a genuinely customized media consumption experience.

Still a question remains unanswered: how to interact with news beyond retrieval and comments? I would welcome any thoughts and ideas on this issue. I am almost certain that others have reflected on this issue before I did, so any pointers are also welcome.

Back to un-prefixed learning

I remember having been exposed to eLearning for the first time around 1995, when I started working as a PhD student on a number of projects focusing on teleteaching. A few years later, I got involved in a project making use of a so-called learning platform, which at the core was a document management system. Of course, these were then called knowledge management systems, but I will not concentrate on the (perhaps more philosophical question) whether it is possible to actually manage knowledge.

Of course a large number of projects in the field of eLearning have had their success, being forerunners in making learning media available, combined with synchronous and asynchronous communication tools (chat, internal mail) and tools to customize a student's access to learning media. At some point, modularization of learning media and tools to combine previously stored chunks of material for students became an important issue, as well as being able to annotate learning material. As new terms for learning support were coined, I remember having spoken at a conference on computer-supported cooperative learning (CSCL). I won't go into too much detail, but every now and then I find expressions such as ubiquitous and mobile learning being employed. Surely technological progress will also result in new application scenarios, but are they really so new - or shouldn't we focus more on what learning is all about from a pedagogical point of view and critically evaluate on what will help to improve learning processes and curricula? Learning programs that help to prepare students better for their future jobs (for those that are not necessarily planning to pursue an academic career) as well as advanced training programs are being installed, but there still seems to be a gap between learning on the job and learning off the job. Thus, the challenge seems to be to embed learning processes in everyday work, thus focusing less on learning in university settings and more on work-related scenarios.

All in all, learning seems to be focused way to much on the individual (independently of the used tools), and not enough on team and collaborative efforts. My impression is that tools will not help much to change this. We produce lots of documents, save them on file systems, web servers or other repositories, but who actually reads them? Who and what might help us to improve the organization of information? Why do we believe so much in tools and the promises that are being made that new tools will solve all our problems? eLearning is focused way to much on tools and processes, forgetting about the involved persons applying their knowledge to everyday's needs. Let us get back to learning and accept that the promises of eLearning were often not based on solid ground.

Microblogging - a quick review

First of all, I would like to mention that my estimation of microblogging revolves around my experiences with Twitter. Many people have written about what they use Twitter for, so I think the time has come that I contribute my own 2 ct. worth.

I think it was June 2007 when I started trying this out - announcing new blog posts, some events and interesting material I was reading -, but already one month later I only contributed tweets very infrequently. Maybe it was because of lack of feedback or interesting twitters to follow, I am not really sure.

Well, at the beginning of this year I started again, and I realized that it can be beneficial to review the list of people to follow. In that context, I found the following classes of utterances (roughly corresponding to classes of users):

1. tweets describing what people are doing (going to lunch, preparing a meeting, spending time with their kids etc.)
   
2. tweets referencing interesting articles to read
   
3. tweets embedded in bilateral conversations about various issues
   
4. news tweets from specific aggregators
   

Of course there are other utterances not fitting in any of these "classes", but what I can say for now is that while I am interested in (2) and (4), (1) is noise to me, and (3) is hard to follow if I am not involved.

As far as my own tweets are concerned, these focus around thoughts regarding newly discovered tools or contributions just read, or references to interesting articles.

As the nature of twitter is unidirectional utterances rather than real conversations, I sometimes feel a bit lost when I do not receive any feedback from what I contributed, but maybe this is because I do not follow enough interesting users.

Mobile Tagging

This is no new topic, actually, as mobile tagging has been around for some years now. I am currently reading an interesting study issued by the German consulting company DETECON (in English).

After explaining the current state of tagging, two classes of scenarios are discussed, namely pull tagging (the user actively focuses on a tag with his mobile phone to retrieve some additional information or to execute a related activity), and push tagging (mobile codes are sent to the mobile device via SMS or MMS).

In the second case, in order to avoid unwanted tags being sent to the user, it seems clear to me that some kind of profiling is needed in order to take into accout the user's current context (e. g. location), or his interests (e. g. deduced from the history of previously retrieved tags). Otherwise, barcodes will be experienced as pure advertisements and not be taken into account by end users.

Finding interesting followers ...

Well, it's been some time I last used twitter - I do have a couple of followees, but how would I be able to find new ones? First thing that comes to mind is to look in other social networks, but I really would like to have a smarter way to do so.

The problem itself is that twitter by itself does not have any intelligent means to classify its users. OK, it is not a problem, because that is not what it was designed for in the first place. From a contribution of 140 characters, you would not really expect any semantics. However, chances are that twitter is used to notify other users about some interesting posts that are located elsewere, referenced through a so-called tinyurl. So, when following these and parsing the textual content, it might be possible to extract some profile assuming that what you read or write pretty well characterizes your areas of interest.

The task of extracting some information from web pages surely is not new, and I am definitely interested in knowing whether there are some available (web) services, preferably open source, that provide a means to extract information from text, perhaps as a tag cloud.

Given that, it is still necessary to add some semantics to user tag clouds, or to at least have some metrics available that allow to compare tag clouds in order to be able to recommend similar twitter users. Let's see if anyone out there has some useful hints for me.

Private & public communication

Maintaining a weblog with the ability to comment on postings can be interesting and fruitful in terms of exchanging information or starting interesting (asynchronous) discussions. However, as comments can be anything, this function can also be misused by some to start bashing people. Then, the question the blog author is faced with is how to deal with such comments that are off-topic and do not have any benefit. Most of the blog providers thus allow three options to handle comments: not allow comments at all (which turns the weblog into a diary and prevents any exchange), allow comments only after review (the comment is only readable by the blog author until being accepted as publicly acceptable), and allow comments without review.

I personally prefer allowing comments after reviewing them, because I am then always in control of what is being commented. After all, this is my blog and not a public journal. An analogy that is often mentioned is being able to decide whom to allow to enter one's home. Still, one could well ask whether this already constitutes some kind of censorship. Assuming that censorship is understood as suppression of published material by someone other than the author or editor (e. g. a state or an influential organization), I do not think so. As any journal or newspaper is able (via their editorial board) to decide what is being published (including any submissions also known as letters to the editor), so should blog authors. If a weblog is established with several authors (e. g. corporate blog), these should of course agree on some guidelines in order to avoid mutual deletion of postings.

Another personal argument on reviewing comments prior to their publicaton is that it avoids any public discussions on comments. If I do not accept a comment by someone, this person will always be able to discuss this with me, without the public participating. On the other hand, if allowing any comments, this may fuel mutual bashing, which is hardly acceptable, and results in the necessity to delete comments by others, which I do not really consider a good thing to do. It is like having to extinct a fire that could have been avoided in the first place.

Searching for you ...

OK, this is not exactly about collaboration, but instead about finding old friends. More precisely, I would like to get back in touch with those I graduated with back in 1987. If you happen to read this, please consider registering to this DFG Alumni site so we can get back in touch. Thank you!

View my page on DFG Alumni 1987

Restricted social networks

I am not aware of how many social network services exist, but I am sure it must be hundreds if not more. Most of them operate on a "come in and find out" basis, which means that you sign up and have a look, and most often, you just stay registered, whether you really use the service or not.

Some of the services are restricted, i. e. they work on an invite-only basis. This may have the advantage that the newly registered user at least has one connection within the network if this person is invited by someone they know, but if any registered user is able to invite any other user, I do not understand what the benefit would be.

I have the impression that the networks that may prove to be the most useful to me are those where I can define what class of relationship links me to the person in question (e. g. family, friend, acquaintance, virtual contact), as it may be employed by application services built on top of a network managing service, e. g. in order to provide useful recommendations.

Business models for e-news

As a recent study entitled The medium is the message by PriceWaterhouseCoopers shows, consumers are not prepared to pay more than half the sum they would pay for a printed magazine. So, if the printed magazine already contains a great number of advertisements to cover the publishing costs, what kind of business models do we expect for online publications if the prize to be charged is low to non-existent? If many consumers see digital-only content as a substitute for printed content, it is hard to argue that the online version of a publication will create only minor additional costs, thus it can live without ads.

The question is what kind of advertisements can be offered to customers so they are not bothered or annoyed, but find them helpful and a hint to useful resources or services - or how to bundle media consumption with other services (such as an internet connection or IPTV service with additional value), thus charging the end user a competitive fee so the included media delivery costs are not experienced as an additional burden. What would an additional value that customers are prepared to value appropriately be like if publishers in Britain and North America expect to generate as much as 20% of their total revenues from digital platforms within the next five years, while in Europe, it is only 10%?

An interesting idea seems to be a stronger interaction between official media providers and blog articles, but this needs a bit more thinking.

Personalized News

These days, telecom operators can be seen experimenting with customized news delivery. While Amazon already offers a portable device for e-books called Kindle (not available in Europe yet), in order to trigger their e-book sales (available at $ 9.99 per item), there are a couple of projects in Europe that aim to deliver electronic news to the customer in a mobile device (such as Orange / France Telecom, which is cooperating with five major newspapers in France in their Read&Go trial with 150 testers, or Deutsche Telekom in Germany, planning a trial under the (project) label News4Me with a few dozen test users in Berlin this fall. The claim is, of course, to offer personalized news, but this requires a cooperation with news agencies or newspapers, as well as a hardware manufacturer offering suitable devices that are portable enough to be carried around, yet offer an appropriate screen size and resolution.

When it comes to news aggregation, Google does a good job when being online - the main feature being the search field to filter out relevant articles. However, when it comes to mobile usage, the following features seem to be mandatory in order to guarantee ease of use and offer enough attractiveness for potential customers:

• a categorization for articles (which could be selectable via tabs), this requires a categorization mechanism, mapping the categories from news providers to a built-in categorization scheme (including the treatment of synonyms)
  
• a keyword extraction mechanism that generates tags for news articles based on the contained text
  
• a rating mechanism, allowing users to influence what is being proposed to them (i. e. dynamic profile adaptation). This could be explicit rating as well as implicit mechanisms (e. g. for articles spanning several pages, if the next page is selected, this could indicate an interest for the selected article). Explicit rating mechanisms should have a stronger influence than implicit ratings.
  
• a profile management mechanism, allowing for an easy generation of initial profiles (e. g. pre-selection of genres) and an adaptation of profiles based on ratings and other user interactions (causing positive and negative weighing of tags and genres to influence the current profile)
  
• annotation mechanisms that assume pen-based interaction, as (virtual) keyboards are not an appropriate interaction mechanism for mobile devices. This includes handwriting recognition and pre-defined labelings (such as indicating consent, disagreement etc.)
  
• forwarding mechanisms to other users participating in the service, including a text field for personal notes
  
• a Web version of the personalized news portal, allowing an alternative usage via ordinary PCs or notebooks, and also allowing the inviation of users not yet participating in the service (by sending them an e-mail or, alternatively, a SMS on their mobile phones)
  

Of course these are just a few thoughts that are by no means complete - any additional comments are welcome.

Self-regulating social networks

The basic idea behind social networks such as LinkedIn or MySpace is that you have a set of registered users you are connected to, but the influence this has on the kind of resources you get recommended is almost non-existent. The most I could think of right know is that your connections to other members are exploited for collaborative filtering. What is missing is some kind of social network model which is able to express the kind of connection that you have to someone else - let's say, someone that is working on exactly the same problem as you are, another person that you usually go jogging with, maybe a third person that you share your love for music with, and the like. It would not be much effort to specify the kind of connection to a person when you ask them to be added to your list, but the requirement I would have is that this relation will adapt itself in conformance to the interactions with the system where you have your social network. In that case, I imagine it would be really fun discovering new people that otherwise you would never have met, or get recommended really useful resources to look at.

Two dot oh

I am getting a bit tired of reading all this stuff about Web 2.0, Enterprise 2.0 and the like. Seems as if you could take any term, add some versioning number greater than two and then you have a new buzzword that you can explain anyway you like.

But this blog is not here to just rant along, so I try to be a bit more precise. The versioning number 2.0 seems to indicate some level of user participation so operators of backend systems (such as a CRM module) get regular feedback from end users as a by-product from user interaction. If you find 3.0 as a suffix, this seems to indicate that the collected data from user interactions are not only grouped and associated to specific instances (such as database objects and processes), but that the relation between the user interactions is also exploited semantically to deduce semantics, meaning and, in the long run, user intentions.

When you read the press, all kinds of neat scenarios are described, but we almost never read how much modelling effort is needed. On the other hand, while being placed on hold after having specified what type of service I could possibly want via an automated agent, I ask myself if I would not prefer to talk to a human instead of a machine. Why do I need to tell the agent that my phone connection is broken if it is possible to check that automatically? Why do I need to spend a minute or more interacting with a machine if waiting intervals are not reduced?

Open Access

It's nothing new: subscriptions to journals are getting increasingly expensive, while in theory anyone can publish anything online. Some publications like the Journal of Universal Computer Science have been forerunners, making available all research papers for free, at least regarding the online version. The more research results are visible to a greater public free of charge the better the advancement in science, at least in theory. This is also good for private research institutions and companies who are interested in cooperating with universities and turning research results into products. As the initiators of the open access initiative say:

The main advantages of open access are the increased visibility and thus the increased impact of scientific articles.

It is not clear what the added value of publishing houses is anyway, as neither authors nor reviewers are getting paid for their job. Publishing online does not really generate high costs beyond setting up a web server with a database and someone that implements user interface and server functions - at least it is a task that only needs to be done once in a while and could be added to subscription costs for bound volumes.

While the costs for subscriptions have doubled between 1986 and 2000, the accumulated inflation rate would be around 40 percent - and it is well known that the budget that is available to universities is constrained. We're talking about yearly costs of between 900 and 3.400 $, as a recent periodicals price survey reveals. When looking at the 10 most expensive journals from Elsevier, who has a share of 13.6%, all of them have a price tag of more than 10.000 $. Who is able to pay that?

The argument goes that publishing will produce one-time costs that may be covered by the authors themselves or their sponsors, but once these costs are covered, the availability should be free of charge.

At the moment, only about 10 to 15 percent are available without constraints - and wouldn't it be nice to see this percentage increase to something like 50 percent? The other question probably is: who will read all these publications, and how will it possible to find more easily what is really relevant for me? This also would require some rating and annotation mechanism in order to make published results more useful to their readers, such as with services like Connotea or Faculty of 1000 for biologists.

Social network aggregation

It would be really cool if there were only one social network to belong to, with the possiblity to tag the kind of relationship to the contacts one has. This is why we now have social network aggregators, which is a good idea to begin with. However, if we think about the content our contacts produce, of which not all would be relevant, given a specific context, I start to think that some intelligent filtering on top of contact aggregation would be helpful. This might also help discovering new, potentially interesting, contacts relative to what my current concerns are.

Tagging is the first idea that comes to mind, however, as any string of characters can be a tag, I feel that some semantic support is necessary. What I am thinking of is some controlled hierarchy with additional cross-links between terms, so any given term would have potentially several paths to some root node. Nothing new here.

When it comes to some well-defined contribution (e. g. a video about a place, a review of some restaurant), this may be not too difficult to integrate, but how about short utterances, e. g. via twitter? If this utterance contains a link, it might be not too difficult if it is possible to extract the kind of relationship by looking at the destination. But how about if the utterance is just some kind of statement like "I just can't stand the weather", or if one utterance is a follow-up of another?

I would be interested in any kind of thoughts regarding the semantic integration of micro-blogging - feel free to comment or to directly get in touch with me! Thanks in advance.

Virtual worlds - do they help in collaboration?

Today I found an announcement in my mailbox that Google has released Lively where you can create a virtual alter ego (also called avatar) together with all sorts of apparel and virtual spaces. Well, after having played around with it a bit (it's free of charge, being paid by advertisements), I can say that, while it is easy to use, it needs a bit of time to get ready to go, without me really being able to see what the benefit is. Maybe one thing that is missing is a number of templates, involving rooms, avatars, furniture, clothing and the like which would get you started really quickly. But on the other hand, I may be one of these persons that is getting too old in order to really grasp the benefits and perhaps the fun that may be involved in using this kind of virtual space.

But the question was whether virtual worlds may help in collaboration, that is, in a professional context. If I do not see the person I am working with (for example in international project work), I personally do not mind, as I am likely to either know who that person is, or, if not, I am able to at least see a picture via their web page or get some other information via their blog etc. What counts for me is to have an impression how well I can work with that person, and on what basis. In other words, I need to know what language (technical, marketing, research, adminstration) my partner is used to, added by how well I can get along with that person (although personal sympathy should probably not have too much of an influence). So, why do I need the support of virtual gadgets? After all, I am not earning my money fooling around with cool stuff, but working on my projects and getting my tasks done.

Personalization and Context

Well, it's been some time that I last posted here. While some say that blogging needs to occur at least a couple of times a week, I take the liberty to post whenever I think I have something to say. This time, it's about personalization, a term that seems kind of fuzzy, as it seems to refer to services that are tailored to the current needs of their users. These needs are related to the current user context (e. g., location, time and wheather), as well as their areas of interest.

To take the example of someone looking for a restaurant, their context will be defined by where they are and what time it is (e. g. looking for lunch as opposed to having dinner), while their area of interest may be partially determined by the kind of food they like.

Personalization and mobility often go hand in hand, as it is rather tedious to enter terms in search fields of mobile browsers. As the needs and tastes of someone are rather not too dynamic, it makes sense to manage them via user profiles, as opposed to context information, which may vary in case of someone that often changes locations.

This said, I am interested in discovering context models and research pointing in that direction. I do explicitly welcome any insights and hints that may help me understand the issue of personalization a bit better.