Does Folksonomy Work for PLM?

First of all, I think we need to clarify terms. Folksonomy is the term used for the classification of mechanisms adopted in social and collaborative environments. This is also known as collaborative tagging, social classification and some other buzzwords you can easily Google. I think that folksonomy is very undervalued by today’s Product Lifecycle Management and Product Development tools. Traditionally, the terms used were classification or taxonomy. Regarding the naming itself, I cannot determine which sounds more complicated – Social Classification vs. Folksonomy – I think both names are bad for end users and tags are much better 🙂

 The main difference between Taxonomy and Folksonomy is the absence of a predefined schema. This is functionality absolutely essential in today’s PLM. In this dynamic world, you cannot leave classification to the administrators of IT /PLM – instead, you need to create your own classification schemas into the way that you work. Since you work in a much more personal way these days, your ability to adapt your environment is next to mandatory. In addition, social classification can be the outcome of your collaborative environment and help you to navigate your daily tasks and activities. You are probably familiar with Folksonomy in terms of a tag cloud.

A data cloud is normally generated from words or tags related to the content you are working on. But you can control this cloud if you have the capability to add tags.

 To be more practical, how can you start using Folksonomy in our current environment? The beauty of all Web 2.0 related capabilities is that you can simply borrow them from the outside rather than having to ask your PLM vendor to develop them. What practical benefits you can gain from Folksonomy? In my opinion, there are multiple scenarios:

 1.     You can allow to users to add additional tags to objects in the system. This is a free rather than a controlled mechanism. Created tags can provide an ideal customizable environment for navigating between objects and using them for contextual selection.

 2.     You can generate tags out of multiple attribute values such has component characteristics, projects, suppliers etc. Since all these values may be not controlled by the administrator, you will gain an absolute custom representation of values reflecting your system

 Now, how can we implement it? You actually have multiple choices. If your system is completely Web- based, and you can address your PLM content (in simple words, objects) via static URLs, for tagging, you can use the capabilities of available social software,  (providing that they exist within your organization). Or, if your organization allows you, you can use public tagging services (i.e. http://del.isio.us) or, if you have WSS 3.0 or MOSS – use tagging web part for SharePoint

 If you are not running a Web-based system, you are more limited and need to rely on API and customization capabilities of your PLM/PDM system. But if you have customization capabilities that allow you to add Web browser components, you can easily deploy tagging components from any open source or other component providers. And, finally you can use some of already available development of CAD/PLM vendors – SolidWorks Labs Tagger. I think Folksonomy is nice and useful, and  need to be adopted soon by CAD/PLM vendors.

 And, finally if you don’t see yourself starting to explore all these capabilities now, you can simply enjoy this video that will help you to think about the practical use of Folksonomy in your life 🙂 –

 

 

Share

Share This Post

  • Hello Oleg-

    Prior to SolidWorks Lab Tag Application, the Tag function in SolidWorks was ‘broken’ or at least ‘half-baked’, see my blog http://designsmarter.typepad.com/devonsowell/come_out_come_out_where_ever_you_are/

    Over and over, we’re told that SolidWorks Lab applications are ‘beta’ and are not recommended for ‘production’

    Any comments?

    Thanks,
    Devon T. Sowell
    http://www.3-ddesignsolutions.com

  • yml

    This is a very interesting post for me since I am in the process of adding this particular feature to a CMS which is going to be Open Source in 2 weeks, watch my blog if you want more information about this. Adding this particular feature to an application is just a matter of days and this includes the automatic tagging using Open Calais API (http://www.opencalais.com/).

    Open Calais is a great resource to tag a piece of text what would be more interesting for a company like DS. Is to provide a service to tag PLM information. Which is a bit different from a news article or a blog post. Of course then user can modify (add/remove) automatic tags.

    This is just my 2 cts
    –yml

  • yml, thanks! Actually I’ve been looking on http://www.opencalais.com/. They are really interesting with their ability to classify and create additional metadata. What source of data your provide to opencalais in your implementation? Thanks. Oleg.

  • Michael Reitman

    Interesting point you touched here.

    It seems to me there are 2 conceptual levels for “taxonomy vs folksonomy” discussion.

    The higher, application level is being discussed in the post.
    It has immediate practical potential and I agree that will see its increased implementation in PLM area in close future.

    The lower level is more of theoretical discussion, at least, at this stage. It is “conceptual taxonomy” of PLM (or other) system – basic concepts, objects, semantics and relationships that system is built upon, foundation of its “data model”. Also defines the “relevance” of the system for certain “knowledge domain” or area of human activity.

    For example, I see a lot of similar efforts in completely different domains to create “knowledge management” systems – from shape recognition to manufacturability wizard to medical diagnostic advisor. Each starts with introduction of semantics and tags to allow building “ontology” to allow capturing information in a knowledge domain and classifying it for further management. And each argues that their ontology (or approach to build it) is more accurate and complete, and would serve as better foundation for future systems.

    While knowledge domains can be pretty wide (Product Lifecycle is one example), the limitations introduced by pre-defined semantics and ontology (essential taxonomy) are hard (or impossible) to break.

    There is chance that efforts invested into research and development of knowledge management systems can bring to reality first “essentially flexible” systems (essential folksonomy) that can be easily customized for the use in different “knowledge domains”.
    Then, hopefully, we may find components of PLM environment (e.g., used for manufacturability wizard) that are “general purpose knowledge management component” customized for specific usage, rather than special application implemented specifically for PLM domain.

  • yml

    Oleg,
    In our implementation we are sending both text or xhtml to Open Calais.
    This kind of technology could be used to classify change management process.
    –yml

  • yml

    oups, forget to paste this in my previous comment : http://www.opencalais.com/Gnosis
    This is a nice way to play around with Open Calais and get an idea of what you can do with it.
    –yml

  • yml, I agree, gnosis is pretty nice. I played with this couple of month ago. it was good for analyzing of plain text. What are principals of classification process you mentioned above? Is it about to extract relevant text pieces from change requests?

    Regards.Oleg
    PS. BTW – is there any relation between ClearForest Gnosis solution and Israeli companies ClearForest? It looks like they are doing the same and belongs to Reuter…

  • To Michael:

    Michael, I agree. There are multiple application of classification. You got right my main point. I think today we can improve behavior of PLM (or any other TLA based system) involved into product development for manufacturing by applying folksonomy based classification. This is mainly because there high level of diversification of processes almost as soon as you are going down from top OEMs (Tier1 and lower).

    What you mention about domain classifications is very useful for KBE (Knowledge Based Engineering) and sometime it crossed with Configuration. Companies dealing with highly complicated CTO (configuration to order) can use it as well…

    Thanks for joining discussion!
    Oleg

  • Devon, sorry about some delay (spam filter captured this). I’m planning to try SW Tags from labs. I haven’t heard about customer experience with this. But it looks nice and I got few positive comments on this approach. You can also see my later post about tagging in general. http://plmtwine.com/2009/04/09/how-tagging-can-prevent-plm-from-a-compulsive-obsessive-disorder-problem/
    -Oleg.

  • Javid

    Hi,
    I’d like to know if there is any way to create atrribute block that can be sensitive to x,y,z cordinates in Autocad.
    that means when i inseert the block into my drawing, it can get the insertion point ID and put it in the value of my block.
    hope i could explain my problem such enough and I’ll be thankful if you help me to find the right way to create this.
    Thanks in advance and geer

  • Javid, I think, you can write a custom command that will do so. You’ll replace block attributes with values of x, y, z you extract during insertion time. Does it make sense to you? Best, Oleg