When you gaze at Sharepoint as of right now … which horse do you see? Secretariat or the Trojan Horse? I guess it depends what side of the walls of Troy you sit on … but lets ponder a bit anyway.
| Webcast Details |
Notable Points |
| Catalyst Article:
Writer: John Fontana
- Senior Editor @ Network World. Lives in Denver. Seems to cover a broad space (Infrastructure) so hard to tell his bias if any.
Read Article? No
- We all know there is a ramp up in Sharepoint sales but what does it really mean in context. That is what I pondered as I stumbled across this article.
|
1. Sharepoint as Secretariat
Being a runner, I get misty just watching the raw power of Secretariat on film. That horse had to love running. Such power. Is MSFT’s Sharepoint similar? The article makes Sharepoint sound like it. To the point you aren’t awarding the triple crown … you do it twice - collaboration, portal, search, enterprise content management (ECM), business process management and business intelligence. Just like “Big Red” MSFT is trying to make a stance that the sum of their parts is better than other combinations (or niche plays). But is that relevant here? Do you need the parts fused? Does that drive notable business value? If it were possible to have such a seamless environment it would be a a notable impact to most businesses … HOWEVER … while Sharepoint has a compelling story here I am still leary if this Sharepoint angle is the only and/or best. Let’s take a few examples.
2. Collaboration - What Working? Blogs, Wikis, Dashboards
When you study the landscape of multiple companies you see that the efficient collaborators are really driven by a few features. They document most of their work in Wiki’s. They share thoughts and insights internally/externally via blogs and then they provide a summary view of that activity from a dashboard. These are all things Sharepoint can do (but so can any coding environment) but not so good in its vanilla form. Personally, Fujitsu researched a variety of 3rd party WIKI solutions for Sharepoint due to their gap in functionality. Same story with Blogs. So in this area the rollup of Sharepoint is really a Trojan Horse in my view. It isn’t the best solution out in the market and the integration in other areas doesn’t trump it. You would be better off building a solution on Google (checkout the early release of Google Sites, Jive Software, or Confluence for example)
3. Portal. Sizzle or Steak?
The Business Data Catalog (BDC) webparts in sharepoint is an excellent example of how MSFT is positioning portal. You get a webpart and point these wizbangs at a data source and wham - instant access. Not so fast McFly. In tactical deployments right now it isn’t working so hot. Most of the time you have to rebuild the BDC connections from scratch getting very close to a service feeding a datagrid (translation its just like custom development in effort and cost). Still, for the delivery of admin screens and simple transactions on custom solutions I do think the Sharepoint listparts can expedite delivery and work well. So I will say this is one area that does intrigue me about Sharepoint still. Thier listparts and BDCs hold promise. More horseflesh than wood here.
4. ECM/BPM - Does Storing Documents Make You A ECM Player?
If you are trying to manage your digital assets (ECM), Sharepoint is an option but in reality there are strong reasons to think harder before leaping here. For starters ask yourself what is the INDUSTRY standard of BPM features. You have on demand solutions like Salesforce.com, Open Source Alternatives like Alfresco, and emerging angles like AMZN’s S3 that (with some investment) could evolve into far better solutions for managing your data (which is one of the core assets you own as a company) while costing you much less money (not a lock in implication). If you are still intrigued by the MSFT offering … move slowly. It would be better to spend your investment dollars right now on how to tag and structure the meta of your unstructured data right (the true value, getting search on documentation going) than choosing a repository to put it in (you have survived this long … right?).
Relative to BPM, there is a solid workflow in Sharepoint that can drive your business but … similar to ECM … I think you have to decide at a macro level around BPM and not just do a rollup decision. Workflow automation (manual or systemic) is a core aspect of your business. There are reasons why major products like Feugo and Interstage have been investing in the space for years. A better plan here would be to understand the macro strategy around BPM for your org in 3-5 years and see if MSFT’s roadmap fits well enough.
Both these aspects are Trojan horses in my view (although I am a fan of the workflow in Sharepoint and MSFT as a whole). |
(** I skipped BI above on purpose as the BDC, to me is the primary unique value statement there **)
In short, I think the core of Sharepoint is about team collaboration and in reality there are better cheaper answers in the market. Study what true collaboration enablement is in your ORG then jump for the shiny widget (which may be MSFT … but may not).
What do you think?
** START OF RAW SCRIBBLE TAKEN WHILE RUNNING **
– No running notes today … spent too much time typing off this article (have to move on the inspiration)
** END RAW SCRIBBLE TAKEN WHILE RUNNING **
March 28, 2008
Posted by
bauertim |
1-Definitely Watch This |
BPM, ECM, Sharepoint, Search, Collaboration, Feugo, Interstage, Jive, S3, Alfresco, Confluence |
2 Comments
Quicksilver. Sounds like we are vampire hunting. Actually it is a prototype for a universal interaction model @ Google based on existing UI concepts of Mac’s finder + command line.
What stuck with me after pondering it awhile is the speed (as you watch it) of its interaction. It enables you to scan through all kinds of digital media, drilling down into it, and then being able to take any relevant actions … all without leaving the UI. In effect a generic UI far more powerful than file explorer for all applications.
| Webcast Details |
Notable Points |
| Title/Link: Quicksilver: Universal Access & Action
Speaker: Nicholas Jitkoff
User Interface Designer at Google.
Recommend to Watch? Yes.
- There are some good concept here around design constructs.
|
1. Designing Interaction to Mirror Speech? - Noun / Verb
It was intriguing to consider Nicholas’ stance that UI should be based on verbal context. Give an icon for a noun on the left side. Then dynamically render relevant actions (verbs) with associated icons to the right. For example, say you went with the noun of MSFT Word — then you picked (from the dynamic set of actions allowed with that noun) the verb of file open. Wham. You then in one click launch MSFT Word with that document. “Thats just like double click in explorer” you say. True, but the power is that ANY verb supported by the noun is possible in a similar efficiency not just a default double click = open. Think actions like = print, merge, compare, email, etc. That said most of the examples in the video are around iTunes (e.g. noun genre - verb play).
2. Forget “Ribbon” or Menu Fast Path - Noun/Verb?
If you hate how MSFT Office 2007 has broken all your fast path keys (unless you remember them by heart) raise your hand. [Tim raises both of his hands]. Not sure why MSFT felt they couldn’t enable both user personas in Office (power user and layman) but they didn’t. Instead they sacrificed one group (power users and their fast path keys) for another. In hindsight, however, if you had quicksilver the power user could have their interaction style (key driven via noun/verb in an office application) while those that don’t know what they want can hunt and peck around in the static ribbon and it’s hierarchy. So in any app you just hit some combo key (ALT-Q?) and Quicksilver would launch with the Noun of that app. Then you get the search with real time results feel like the start menu in Vista but with the possible verbs in that app. If I could do that I might forgive MSFT for their Ribbon.
3. Constellation .. Visual Drill Down?
Another interesting UI concept shown by Nicholas was creating (1) showing noun/verb as a circle with the verbs like petals off the core of noun and then (2) allowing the user to drill into petals (verbs) which then created a new noun/verb layer over the existing one. What was interesting about this is that is allows users to keep a much more visual context in place of where they came from and the ability to roll back in the layers of noun-verb very quickly. Far better than todays model of breadcrumb hierarchy representations (think Vista’s file explorer model). I have also seen a similar UI concept on Doug McCune’s excellent Flex UI model called “Tile UIDesktop”. About 45 seconds into his 3 minute demo of that he starts showing how his implementation of ‘petal’ actions around a selected object on his UIdestop (the ‘noun’) works. Really gives you a sense of the smooth interaction of the noun/verb concept in a petal format could work. The true efficiency in his is around the fact that when you right click (I assume that is the event) to expose the available petals (verb) for the selected object (noun) it presents itself right where your mouse is. Classic Mac design … no extra mouse movements. Just a slight flick of your mouse, click, and go. Very nice. |
In short, this Interaction model of Noun/Verb coupled with UI constructs like petals (e.g. McCune’s example) in my mind could quickly trump classical menu path models. Why have a fixed menu structure. Let it be dynamic (if no noun/object is selected allow it to be anything like quicksilver). Here are some images of quicksilver / constellation.
What do you think?
** START OF RAW SCRIBBLE TAKEN WHILE RUNNING **
Quicksilver
GoogleVideo
–File finder
–Extra fields
–Play it
–Left side is app
–Right side is actions on the app
–Sometimes 3
–
–Noun, verb, indirect object …
–
–Search, Summon (search for object and open), browse (hiearchy, search to hierarchy node then drill), acting (file–>compress–>new object)
–
–
–Fast universal access (like a google)
–Not just files/web pages/contacts/people
–where data is (local, cloud, etc)
–index backup / stickit
–search this page … my content .. jump to it
–
–ignore boundries of apps
–app package is meaningless … abstract file system … case by case by apps … show just mp3s ..
–
–Acts w/o doing
–Way w/ applications (talk)
–
–Find something and get there
–not complex apps (like photoshop)
–Simple apps like email
–automater …builds of chains of actions
–
–
–services layer … apple … put in to allow manipulate of anycontent
–
–
–enzo launcher for windows
–ask for word … define
–override ..
–context … makes flexibility
–
–
–add todo
–get milk
–to file
–dont leave the app … do in context
–quicksilver is transparent
–act w/o doing
–
–Act w/o doing
–Effortless flow of how the systems work
–
–
–How fast you can go
–limit on search isn’t optimization
–limit will become users speed of typing
–worry about what is the right thing
–abbreviation mashing
–dont type full spaces … prefix … apho .. gets photoshop
–it learns
–apho goes w/ you … your TLAs
–
–
–How much stuff
–Get at data and functionality
–Growing
–airport locations for example is mixed in
–control and manipulate the data .. pipe between apps
–initial version building
–ton of code per app interfaces… shell scripts … app
–move to plug in
–specific app … actions …
–Actions come from plugin
–create new objects by actions
–
–actions are objects
–noun verbs are same
–
–
—magic and mystery … UI blog … mike krymenski …
–make it seems like it just works
–alchemy … interesting way to go
–Abracadabra & Constellations … CMU last year … students … simple gesture system .. not intelligent .. watch draw .. match to pattern
–mouse based interaction
–constellation (show the things and navigate) … like mccune .. better in that he allowed layers of circles .. drill .. new layer on top
–explore it
–
–
–build into quicksilver
–constellation in was hard … to quicksilver
–new framework called alchemey .. access plugins .. exist with it
–when you are programming
–clutter
–letting people do search for actions
–actions ..specify a noun .. specify current context as the noun
–
–
–Google
–Interconnect the apps
–Email into GMail
–Very hardcoded
–Google could build a suite around this model
–# of search boxes on a machine … all search different … same behavior … contextual search … could tie into universal
–Search contacts and act on them
** END RAW SCRIBBLE TAKEN WHILE RUNNING **
March 27, 2008
Posted by
bauertim |
1-Definitely Watch This |
Apple, Flex, Google, Interaction Design, McCune, Quicksilver, Search, Tile UIDesktop, UX |
No Comments
Interesting, antagonistic, yet somewhat staged keynote @ MSFT’s Mix a few weeks back. They had Kawasaki (an Apple evangelist) interview Steve Ballmer (CEO MSFT) on the state of MSFT. Here was the net of their verbal banter in the first 35-40 minutes.
| Webcast Details |
Notable Points |
| Title: Kawasaki Interviews Ballmer
Speaker: Guy Kawasaki
Variety of notches in Guy’s belt. Head of marketing for Apple on Macintosh product in 80’s. VC now as well as writer. Handful of startups he is driving but none have done much.
Speaker: Steve Ballmer
CEO of MSFT since 2000. Not sure if I would give him a stellar rating in that role over that period — Google taking search space, Apple taking consumer products, Open Source pressing into Enteprise Stack, not gaining foothold in enterprise app market.
Where: URL
Recommend to Watch? Maybe.
- If you like a staged banter, this might amuse.
|
1. MSFT befuddled on how to get share of search market
It was interesting how Steve shifted off Guy’s questioning around the Yahoo bid. To me MSFT seems befuddled on next steps .. leading them to a buy reflex. It may be better for them to innovate a search angle around micro markets like blogs. The blog authors (or similar microsites like social networks) will go w/ a different advertising provider if greater value can be produced than Ad Words. That could take the form of a price war … but hey if they are serious.
2. Recruiting Struggles @ MSFT?
Guess even the big dogs struggle with attracting top talent. Interesting how he points out landing people is rarely about the top compensation package … rather the role and potential to impact. Implies that startups can innovate but can’t reap/scale that innovation. How much do you have the scale an innovation Steve to be successful Steve? Is scale where fulfillment is for hires?
3. Ballmer’s Week: 1 day personal study, 1 day customer, 1 day internal meetings.
This was an interesting point from Steve on how he tries to create balance in his weekly calendar. Stephen Covey would be proud. He has sharpening the saw, relationships, and operations. |
They blathered on around other topics - Vista, Firefox Emergence, Silverlight, etc. But really, the key was watching Ballmer continuing to demonstrate transparency and portraying MSFT as the underdog. I suspect they are concerned with the partner base beginning to shift to other alternatives (SaaS, ERP, Open Source, etc) and are tyring to calm the troops.
As always, here are my raw notes for those amused.
** START OF RAW SCRIBBLE TAKEN WHILE RUNNING **
http://wm.istreamplanet.com/customers/ms/300_microsoft_mix_080306.asx
Kawasaki Interviews Ballmer
*Garage Ventures, Trumors.com
*Ballmer
–Why Yahoo? Search and advertising on internet. Will be next big thing. They need to be in there. They made an offer. Waiting.
–Zero Sum Game requires Killing Google? Yes we have to have a strong position.
–Who is primary competitor? Google? Apple? Four things - desktop (no google, apple yes), server/enterprise (ibm, linux..no google), entertainment/devices (no google has emerging via youtube/etc), online (google dominants .. little engine that could)
–Apple thoughts? Apple does a good job … entertainment / devices … could be losing market … to iphone … desktop
–Guy? Acknowledges need for exchange server w/ mobile … which gives MSFT the enterprise market (better)
–Facebook? An advertising focused partner. Partner with them. Buy small piece of the company. put a few $100M. Ford guy (father works for ford). Few $100M not small. For MSFT w/ $60B in year not huge.
–What drives him? silverlight 2.0 excited. change the world #1. Works w/ the smartest people around. #2. Challenge and competition is good. #3.
–Describe day? Three types. (1) customer meeting day. (2) doctor in office internal days (3) think/write/research day.
–Email. 60-70 pieces. got mail from spam. distro lists abusive not people. no virtual assistant.
–where is bill / post bill msft? this year goes part time. focus on foundation.
–Different way to have whole > sums.
–Recruiting young employees. Best brightest. Work on things that change the world. People need $. Most people think they wouldn’t get $. 1990. Stock was flat. Get rich isn’t primary source of recruiting value.
–Most startups fail.
–Guy’s son won the tournament ..for Halo.
–OS Wars, Anti Trust, etc — no history w/ kids. Young people should know MSFT does more than XBOX. Ballmer –> Opp is same. Make great products. 10-15 years ago what was hot isn’t good now?
–How does guitar hero, xbox, fit in? 17M xbox owners, 11M live subscribers, 1B windows users so opp is there. % of users are free w/ live.
–Silverlight? PC/Internet. Design/develop for mass adoption (HTML) or do RIA. WPF + Silverlight … gets best of PC and WWW in one form as a dev/designer.
–Silverlight launched last year? Last year. 1M downloads a day. Broaden # of desktops w/ silverlight. WPF is high % as part of windows update.
–vista? Heavier than my PC. Heavy. Powerful. Missing drives (talking about AIR). DVDs are passe.
–MSFT lost focus. Move forward or become less relevant. No thing like specialization. Established in desktop/enterprise. Going for two new … consumer (appl) and search ().
–Windows market share. EU.
–Vista. Consumer issues is app compatibility and driver issues. Drivers getting upgraded. SP1 shipped. App compatibility. Uptake will increase. Consumer market.
–Firefox. IE. What is happening there. Built presence and position. Firefox position.
–Browser innovation. Firefox. Safari.
33M in (continue)
** END RAW SCRIBBLE TAKEN WHILE RUNNING **
March 24, 2008
Posted by
bauertim |
3-No Go (unless you are bored) |
MSFT, Recruiting, Search |
3 Comments
It seems nueral networks need to get pulled out of the sock drawer you have your Ton Loc album stashed in. At least if you put any weight into why Google had Geoffrey Hinton in for a talk a few months back.
Being a guy who did his thesis on neural networks (and referenced Hinton quite a bit in the process) I thought it was my duty to watch this discussion, and, as you might expect there were a few salient points:
| Webcast Details |
Notable Points |
Title: The Next Generation of Neural Networks
Speaker: Geoffrey Hinton
Where: URL
Recommend to Watch? No.
|
1. Google’s Interest in AI - Auto Generation of Semantics for the Mountain of Unstructured Data.
At least my guess. At the end of the talk Geoffrey began to apply AI principles against the riddle of tagging massive stores of documents (hello Google) with semantics (hello RDF). It was a good reminder that the best categorizer is the human mind … but when the scope of work is too large it would be better to create a machine (AI) to mimic the brains process of categorization.
2. Unless You Are (1) Google or (2) an AI flunky … only dip your toe.
Again my opinion. AI remains a very complex field so even if you have a project like #1 above I would only dip your toe in this field. As in consider the core conventions that thought leaders like Hinton are leveraging in their models (such as how to evolve a data vector representing a document’s feature set). Then stop. Put down the neural network …and go back into classical rule based approach to AI from there. Better yet, ponder how crowdforce models (getting your employees, customers, etc to do the task in the course of their other tasks) can provide you with a living neural network that categorizes your ‘stuff’. That might be a better riddle to solve. |
Still the topic is interesting and relevant today. Especially as the issue of semantics evolves on digital assets. I will need to dust my thesis “Image Segmentation of the Prostrate Gland via Nueral Networks” off and read it again. Not.
Here are my raw notes for those amused.
** START OF RAW SCRIBBLE TAKEN WHILE RUNNING **
–Perceptron. Didn’t adapt. Learn weights to decision units. Limited.
–2nd Generation. 1985. Feature detectors. Forward / Backward passes.
—Works for non-linear units.
—Huge disappointment.
–1990’s .. Kernal methods tangent … perceptron … training example to a feature .. then optimize off features and weight ones that it keeps … fixed backprop
–Back prop
—Bad as required labelled data
—10^5 is too big … learning time did not scale
—Neural issues
–Learn to do computer graphics (infer vision from there)
–Binary stochastics neurons
–Output of 1/0 … probabilistic
–Learning Module .. Restricte BOltzman Machine
—Can update units in parallel .. no lateral interaction
—Governed by energy function
—energy function drives network going to low energy states
—parameter on connection … vary via feature detectors
—Compare on/off … when fantasizing (no data input) vs data exposed (real) … adjust weights … very simple neuron focused learning .. requires 100 steps … so slow
—Just run 1 step … now fast … change based on statsitics of data and real data
—put random weights on connections, put in data, reconstruct image, then loop (activate, increment, decrement)
–25k connection strengths (each is feature) … connection strength represented by pixel bright …
–Example … so very good at creating twos … give it a two … recreate a two
–Example … data in not trained on … give it a 3 … it pukes 2’s
–Now train layers by features … get more abstract features … go up 3 layers … learned WITHOUT knowing the labels … more layers is a better model of training data
–Visibile vector creating another similar vector
–Find a distribution that is simple than real life data … then find a parametric mapping of a data distribution … in short remove some ‘noise’ … weights evolve in that manner
–his model is 3 layers … 2 layers go back/forth … use for perception imprints … pattern established … then hit the 3rd
–example .. hand written digits … learn 500 features (nuerons) … 28×28 image to 500 … 500 middle … 2000 top layer … 10 label neurons … dont tie label to top … just form loose associations
–top level (2000) is a boltzman machine … low weights … energy landscape w/o view … then cut channels in weights
–learns long / narrow ravines of energy but different enough … to distinguish between 1 and 2’s
–variation of learning passes … tune at end … up pass, multiple on boltzman layer (1000) … then down pass to input one time adjusting up connections
–21:55 — Nice example of perception learning running … shows image of weights changing
–23:40 — Good example of generation …
–Brain state and mental state … neuron state vs perception of your mind
–26:10 — show how the brain state moves between various forms … weird fives …
–27:30 — Comparison chart to other learning models (note K: Nearest neighbor 3.3% is what I used on my thesis)
–discussion search via backprop
—hard to search … get stuck in geo local minimums
—work around by tying to features …
—learn same net w/o labels … put labels … on areas … tune various pieces
–32:00 Could apply to document database. Document vector versus image.
–Finding binary codes for documents
–Use document as a query … create a document vector …
–40:00 Summary — Restricted Boltzmann Machines good for image representation of large bodes of documents
** END RAW SCRIBBLE TAKEN WHILE RUNNING **
March 21, 2008
Posted by
bauertim |
1-Definitely Watch This |
ECM, Search, Google, Neural Networks |
No Comments
Watched this webcast this morning.
Advanced Search in MOSS 2007 — Bill English MVP, MSCE, MCT MindShare
I was just looking for background noise but this one turned out to be quite intriguing. Here are some key thoughts.
1. Value in Knowledge Aggregation I wouldn’t have put together the value of a singular aggregator embedded in the framework for collaboration till I watched this. Consider our Communities of practice. Would love to be able to embed into a give team site (say Project Services) an aggregated search index for ALL sources known to the big guy (intranet, extranet, web, ERP, etc). One stop shop for knowledge. Definite value message to clients … stop surfing the web for tidbits … have it be crawled for you.
2. No MOSS on this Stone. Sharepoint 2003 search was weak … due to indexes by site w/ no cross index relevance. Listening to the improvements this guy rattled off (including one index for all — so relevance will work) made me raise my eyebrow. I think they have really improved thier product here (still has gaps … but doesn’t everyone?)
3. Clarity of Why Help is Needing Setting Up MOSS? It was an interesting point by the speaker. If you deploy a search solution and it doesn’t work well to start … users will stop using it. MOSS can work well if you THINK about it awhile (model how your users will be searching, etc). Clients risk spending a ton of money rolling out content aggregators but not having a solid search design on top of it … and getting no ROI from the investment. We need to help them understand the complexity of setting up search CORRECTLY and play from there.
As always, here are my detail notes. I actually didn’t get to the end of this one (doh!) I suspect he kept going to around 1:20. I included a screen shot of search differences (2003 to MOSS).
| Quote: |
• Bill English MVP, MSCE, MCT MindShare
• Tech Ed (live in Boston)
• Mindsharp
• 2:00 - MOSS / SP 2007
………….○ 50% MOSS
………….○ 50% wss v3
• 4:00 - Search Strat
………….○ Content on desktop / fileservers (unstructured)
………….○ ERP content (Structured)
………….○ Desktop, Team, Division, Enterprise, Extranet, Internet
………….○ In short, crawl structured/unstructured data at any of these levels.
• 7:00 - Search Consistent
………….○ New core search engine is in WSS
………….○ Install MOSS on Top of WSS … add functions for enterprise
………….○ Majority investment in relevance / ranking (#1 customer issue due to multiple index files so no cross ranking across files). Create one index
• 10:00 - New Search Concepts
………….○ Click distance. Lower in result set if authoritative site resource is multiple clicks away. For example MySite (far away), Portal home (good).
……………………..§ Need to edit (in admin) the relevance sites (top, 2nd, 3rd are the groups)
………….○ Anchor Text. URL hyperlink text (on hover) will be indexed.
………….○ URL Depth. Higher in hierarchy rates higher.
………….○ URL Matches. Query matches words in URL name.
………….○ Metadata. Extracts from docs titles / authors .. 1st slide is title author on 1st slide. Not talking about meta fields (it derives it). By file type look in certain locations of file.
………….○ File Type Bias. PPT is more relevant.
………….○ Text. Typical full text.
• 20:00 - Other improvements
………….○ Hit highlights
………….○ Search center
………….○ Duplicate collapse
………….○ Did you mean (google)
………….○ Alerts (result set changes … only once a day or week)
……………………..§ Immediate wasn’t really happening in 2003 (depends on how often you crawled)
• 22:30 - Indexing
………….○ Choose what to crawl, rules, log
………….○ One index per SSP (Shared Svcs Provider). It is a ‘farm service’.
………….○ SSP design is key. One index per SSP.
………….○ Multiple start addresses per content source … so you can crawl segments of a source. Don’t need to create multiple content sources (ie 2003 way).
……………………..§ So you could crawl all the SP blogs on the web under one content source
………….○ Search scopes are from managed properties. In 2003 URN. So you can roll up same ‘content’ fields but separate fields and search it as one field (so multiple system definitions of ‘author’).
• 29:46 - Crawl Process
………….○ Load (protocol, index, site path rules, crawl settings, site hit freq).
………….○ Connect
………….○ Stream meta, content (pass to plug ins - indexer, archival plug in (schema plugin)). No more persistent query service. No category assisstant / topic assistant.
………….○ Meta in sql (not sps.edb)
• 32:20 - Arch Updates
………….○ Alerts daily
………….○ Shadow indexes (resolves network down issues) … continuous propagated to query server (chunking) … scheduling w/ 200-300 content sources is resolved (problems in 2003 at least). More available … quicker to end user (chunking). Query server in MOSS>
• 36:00 - Demo - Creating a Content Source
………….○ Types (sharepoint servers)
………….○ Web sites
………….○ File shares
………….○ Exchange public folders
………….○ Business Data (BDC searching of ERP)
………….○ Q&A
……………………..§ Unique security for above done by site path rule
……………………..§ File extension defaults 2003 and above (pdf filter, and others are still 3rd party)
……………………..§ 2003/2001 .. Add on for notes … in 2007. Don’t know if can crawl in 2007. Assume can via BDC.
………….○ Add site path rule
……………………..§ Hit page but only index below
……………………..§ Crawl complex pages (ie ones w/ question marks).
……………………..§ Sp as http
……………………..§ Add account pw (https certificate)
………….○ Search center
……………………..§ Pick site(s)
……………………..§ Get folder and doc
……………………..§ To customize page (search results).. Can edit page now
……………………..§ Flag ignoring of buss words (like G&R)
……………………..§ Stemming was always on … in 2003 ” was indicating proximity (so you can’t look for exact phrases still)
……………………..§ Q&A
…………………………………□ Will it crawl SP Service? Hit top but wouldn’t go lower in 2003. Not sure in 2007.
…………………………………□ Customize Search Interface? Yes. Just like results page.
• 58:00 - Broader Content Aggregation
………….○ Blogs
………….○ Wiki’s
………….○ ERP systems
………….○ Colleagues and Partners (people search)
……………………..§ Search people and their expertise
……………………..§ Reporting structure
……………………..§ Teams in past
……………………..§ Teams now
……………………..§ Who knows who
……………………..§ External contacts
……………………..§ Connections
• 1:03 - Shared Search Services
………….○ Indexing is resource intensive.
………….○ Avoid redundant
………….○ Always on
………….○ Control central, managed by consumer
………….○ Users can define search scope. Tied to content source. You can search pieces of the index. Scope by any meta data you want (authoer, project, customer, etc)
………….○ 1st impression of search is key [deployment point, weave into discussions w/ clients] … if it isn’t good the 1st or 2nd time they will leave or stop trying … [push UX]
• 1:09 - Demo Create Scope
………….○ Show how fields map to one field (for singular search)
• 1:12 - Query Reporting
………….○ See what people are looking for |
June 27, 2006
Posted by
bauertim |
Uncategorized |
MSFT, Search, Sharepoint |
No Comments
Subscribe
