At the SAP Inside Track event on June 25, 2010, there were *3* tracks/rooms going simultaneously. My choices were partly clear, others required a coin flip or other decision influencer before I went. One with no rivals was Marilyn Pratt, Trevor Carlow and Christine Merton’s Introduction To Knowledge Exploitation, partly because Marilyn was in on it, partly because of her blog (also) titled “Introduction to Knowledge Exploitation” and mainly because we all use and benefit from search every day of our digital lives.
This blog deviates from my usual style, or styles, so bear with me. You will be able to read a summary of several sessions, including primarily one that ran almost an hour. While you’ll read it in a few minutes, it’s taken me a few hours to review the recording, format the content, and discuss meaning and futures.
Jon Reed posted an excellent summary of the event:
|My new PAC blog post: “Analyzing the Success of SAP Inside Tracks.” http://tinyurl.com/2cjkjzz 3:57 PM Jul 1st via web|
(1) Sustainability and Audit
The first session I’m commenting on wasn’t captured; I’ll highlight one part of the conversation – the mashup of SAP Sustainability and Audit. Dennis Howlett attended virtually; coming from an accounting background he had the command and control perspective. I was attending physically, as the speaker Bhavesh Bhagat is the MD-Baltimore ASUG Chapter chair.
What I didn’t know before I sat in was Bhavesh would state that he believes SAP will be moving toward what I’ll call environmental audits, paralleling SOX-type financial audits. I’m not sure the source of this claim, and will be setting up webcasts through ASUG to explore the subject matter, and it certainly pricked up Dennis’ remote antenna ears.
= Deep dive on SAP audit – this is the real fun stuff #sapitnsq 1:56 PM Jun 25th
In my notes, Dennis and I exchanged a couple direct messages on Twitter, which in respect of mutual privacy, I won’t repeat here.
The gist of the exchange was to question Bhavesh for more depth. Alas, the audio is lost, so we’ll get into it another time.
(2) Search and Destroy
As background, please see Marilyn Pratt’s post:
The main event wiki page:
IT206 – Knowledge Exploitation – Getting what you want when you want it – Marilyn Pratt, Trevor Carlow, and Christine Merten
I’ve transcribed chat from the above recording, along with several time stamps, so you can fast forward to the specific material you might be interested in. I’ve also attempted to transcribe verbal comments, with those made near the microphone more likely to be audible than those farther away.
The list of characters below is not complete, either for the 15 or 20 in the room, nor the 15 or so on remotely. These are the ones who texted in the chat, or spoke clearly enough in the room that I recognized them (or because Laure or another transcribed their comments).
I’ve deleted a few chat comments that were not germane to the primary conversation, in the interest of space. Typed-in text chat is generally placed inside tables, while verbal comments are within quotes. The rest are my views. Unless I mangled the transcription.
The chat text, while originally done within the webcast session, sometimes uses “twitter-like” notation where Person A replies to Person B using the “@A” token. People have adopted this slang over the last few years; these messages aren’t from Twitter itself. I wasn’t monitoring Twitter at the time, though you’ll see comments from Jon Reed below in both the chat text and verbally. Laure attempted to repeat what was said in the room to the chat for the benefit of those on the web.
Anyone who’s ever been in a courtroom, and then read the transcript later, knows: “you had to be there.”
|Stephen J||Steven Johannes||@sjohannes|
(13 attendees showed online)
Trevor Carlow / Marilyn Pratt are in the room, talking about topics such as “finding information”, “knowledge exploitation” and asked who in the crowd does this. I believe I answered that we do “Knowledge Management” as part of our day jobs, referring to both creating content and looking for it, perhaps cataloging it, sort of like what I’m doing here.
Trevor showed new selection criteria under development for the SDN portal; I’ve transcribed many of them here:
- Asset Type
- Solution Suite
- Open/resolved/possibly resolved
Those in bold are available now.
Trevor mentioned his background in “Semantic search”, for 2 years he’s been working in this area: products included search, federated search, etc. He said execs (and really all of us) think that “search is a magic wand”, disregarding the serious technology underneath.
SDN has multiple sites (hah), so searches may need to span a variety of systems, storage media and nomenclatures.
8:20 – first evolution, then project “search 2.0”
Now: “Search 3.0”
User identity tailored results
(inaudible) – searches originating externally. Jon Reed had mentioned before the session started that he uses Google to look for content within SDN, feeling that the native search is inadequate, and that his needs are met by using a more familiar search engine interface.
|Chuck Dickson: Can’t hear questions. Could you please repeat?
Laure Cetin: Yes, sure. The question/comment was about SCN members using google search to find content on SCN. Instead of using the SCN search
Trevor explained the added value inside SAP’s search engine, where phrases might be expanded, so “BPX” returns hits with “Business Process Expert”.
Type completion: I think there was a discussion of Google and other applications that do auto-completion for you, in a sense guiding your search to frequently looked-for content, specific spelling, or even (gasp) sponsored content.
David Branan brought up search side effects found on sites like Amazon, where suggestions might be made from your search criteria: “People who searched for _ also searched for _”. This could help “train the user” into choosing search entries in the future, or for the current search, help guide you to what you wanted to find in the first place (after all, it isn’t always in the last place you left it).
|Stephen J: Opt-in for tracking.. i.e. allow vendor XYZ to track your data anonymously to improve user experience such like Microsoft does etc|
16:00 Ratings and reviews
Then, the discussion went into questions, answers, and more about rating content. Trevor started by asking how forum posts, or blogs, might be rated. I said, from the back of the room, that there are ratings on podcasts, but of the several that I’ve uploaded to SDN, I only found one had been rated, and it was given 5 stars by Jon Reed (who was sitting a couple chairs over from me).
The discussion went into the merits of users rating others’ material, such as on Amazon, blogging sites, etc., including star ratings, thumbs up/thumbs down. Note that this discussion is distinct from the SCN points system, as will be evident momentarily when David Branan speaks up.
|Laure Cetin: Jim was mentioning the rating on formal content (articles, downloads, podcast, etc)|
There was a comment about ratings being stratified, such as all high, all low, or mixed high and low with no middle ratings; Jon Reed talked about political material that gathered all low ratings as opponents piled onto anything disagreeable to their views.
|dahowlett: Ratings can be gamed. On ZDN we ditched “thumbs down” because of that|
Jon Reed said he was troubled by ratings.
|Stephen J: @dahowlett +1 -1
Jelena Perfiljeva: I found rating on some eLearning items to be inadequate..
Christine Merten: Jelena, why exactly do you feel ratings are inadequate?
|Stephen J: @dahowlett Yes time is a problem unfortuantely
Tobias Hofmann: why not use a rating system like it is used in the academic world?
Tobias Hofmann: what counts are references from others
Stephen J: Perhaps a more tripadvisor type review/rating system
dahowlett: @tobias – that kind of peer review would feed into some of the ideas we have around certification. To me that’s a good idea that can improve value and quality.
Tobias Hofmann: good content gets a higher rating when it’s getting referenced by other content
David Branan: “what about the existing rating system, the 800 pound gorilla in the room”
Trevor talked about the scores trying to gauge the community rating of an individual
David said there are design flaws in the [SDN] implementation, as ratings are done by item [like a forum answer], and that community reputation eventually accumulates (except for cheaters, who are deleted from the system)
Ratings of forum posts get 3 choices: partial, good, and complete answer.
The web logs rating was graduated, with better posts getting higher scores (and still is, think Marilyn said this), while other content may get flat rates for each contribution.
The “quality is inferred” by the points, at least theoretically, though as you can tell by the chat, that isn’t always the case.
The discussion went into the merits of two parallel systems – a rating system and points system. Management and communication of these distinct avenues is problematic.
|Jon Reed: @tobias yes- linking by other content is a good criteria
dahowlett: @jon @tobias +1
Christine Merten: @tobias – so like google does it – more links to content = higher relevance
Tobias Hofmann: @christine: yes
Stephen J: weblogs are awarded by “experts” not general community
26:53 – You should see a “note pad” of action items on the screen as Trevor began to be bombarded with ideas. I often see this with webcast presentations, as people warm to the subject, and audience comments begin to trigger critical thinking and discussion follow ups.
(“Jim has 20 billion points – we can’t search based on that” – David B.)
|Tobias Hofmann: when someone is writing a blog that can be seen as a “standard” solution: other would link to it in other blogs, forum posts, articles, even in OSS messages
Tobias Hofmann: @stephen: could be any content, even SAP Help content
Stephen J: @tobias I was talking about today.. not future..
Trevor claims: “Google works, because it is paid for, and that Google is the line in the sand: as good as or better, or worse, and terrible.” I think that may be generally true, though users may expect different behavior in site specific searches. The fact the Google (and other search engines) permit selection criteria limiting results to specific sites in a sense dictates that local search results must factor in differentiators that are hidden, or obscure, to general search technology.
29:50 content created by “author” demo
|Stephen J: So when will the Thomas Jung search result page be available 😉|
“guest” came up in too many search results, since that user ID might be the conglomeration of a horde of ex-users. Having this content show up near the top of results has to be frustrating for users, especially as the underlying content is presumed to be dubious to begin with (the accounts may have been suspended for poor manners, or worse).
architecture replaced and repaired – Trevor discusses backend server and application base changes made to fortify the search engineering. This is critical as the content continues to explode.
|Jelena Perfiljeva: Will there be an ability to save search settings or defaults? Current default for the posts created within last 90 days is driving me nuts.|
I asked a comment (in the back of the room), wondering if author search could be expanded to include community credentials such as Mentors, moderators, top contributors, etc. I would not presume that these individuals (myself included) had the best answers, but as long as we’re brainstorming possible system changes, why not? Or, block them from the results, maybe.
|Christine Merten: Question: what do folks think of the filters on the left side? Are these valuable? Do you use them?
Stephen J: I use the filters sometimes when the result list is too large
Jelena Perfiljeva: I think filters are OK, just want to make sure we’ll be able to navigate back to previous results and remove a filter
dahowlett: Those filters look ‘good enough’
Stephen J: oops the new filters look interesting.. but the question states for forum questiosn are not as useful people don’t close questions
Laure Cetin: Comment was to add filter for Mentors, Top Contributors content etc
Stephen J: Can we tag SAP Employees who are Product Managers into a search
Tobias Hofmann: can I combine the filters? Forums + Blogs + Wiki?
Jelena Perfiljeva: Yeah, same question. We have 3 SD forums, Iwant to use all 3 in search.
Christine Merten: @tobias – yes for certain filters such as topic, solution
At some point, we need to limit the number of selection fields, “don’t want to overburden the UI”, not going past a certain density of options for users. Finding that happy medium may take experimentation, usability surveys and polls.
Trevor talked about the underlying architecture; it is now All TRex, while SAP Enterprise Search is being worked on, internal pilots are happening.
A section of the recording was blanked out while Trevor demoed really new cool stuff. Like I said, you had to be there!
Future versions are planned to search into OSS (the SAP Service Marketplace formerly known as the Online Support System), with market content; goal is one-stop shop. This will require a security infrastructure that leverages the customers identity.
41:16 when can we test it?
Trevor, Christine and Marilyn will get back to us.
|Christine Merten: @jelena – are you referring to the forum search on the forum page?
Stephen J: @marilyn +1
Jelena Perfiljeva: Yes, but I think any search does that now
Stephen J: @marilyn +1
dahowlett: Good session +1
Jelena Perfiljeva: You have my name and my email should be on SCN. Please feel free to contact. Thanks!
Stephen J: good session +1
Trevor talked about SDN search working on mobile devices, such as the iPod touch. This is going to be more and more important as the technologies advance.
(3) Recognition and Contributions
Marilyn and I continued the conversation in the back of the room, spilling out into the hallway as we dove into the inequities of the current recognition system, were a few articles or blogs get more weight than dozens of forum posts. We talked about the improvement Laure Cetin is working on, starting with the published “Find The Expert” link. As Marilyn had her camera, she asked if I’d comment on the record.
The video that follows captured part of what I had said, but inevitably, by the time the red light was on, a few thoughts fell to the wayside. Or perhaps, some sentiments are better left off the record.
Thoughts And Concerns
- Continued evolution of search functions in SDN are wanted by the user community.
- Changes in flight will add new categories for filtering search results.
- We want to test them, with representative data, using typical business scenarios, to prove the value to us and to SAP
- Google and other “front end” search engines will continue to be the gold standard of comparison
- It makes sense to complement Google, etc., with features they can’t implement.
- Privacy is a concern. Deal with it.
- Community recognition can be a factor in searches, but should not get in the way.
- Evaluating content by voting through the search engines is a “Fail Whale.”
And now, to Oz (inside joke).