The IBM Visual Recognition classifier is simple to use and works well. However, custom classifier creation is expensive ($0.10/image) and time-consuming. Accidental deleting of a custom classifier puts any workflow using that classifier at risk. There is no obvious way in the API or dashboard to download, duplicate, or lock a custom classifier. This is a concern for production use.
How can I back up a custom classifier created using IBM Watson Visual Recognition? This question went unanswered on IBM's developer forum and I am hoping someone from IBM can provide guidance here.
Thank you!
There is no API method to back up or download a custom classifier. However if you still have the training data you can train a classifier with the same data and you will have the same performance again, with a different classifier_id. This would incur additional expense, as you noted.
We recognize the usefulness of this idea as a feature and will take it into consideration for development.
In the meantime, if a classifier is accidentally deleted or changed, please file a Bluemix support ticket, we may be able to help.
It's currently not possible to backup or download a custom classifier (only create, update, and delete them via the API).
Related
I'm working for a startup and we have a customer who uses Demandware for their online store.
Our company provides vouchers which are worth a specific amount of money. So customers on the retail site can enter the voucher number near the end of the checkout experience and if it's worth, say, 10 dollars, their shopping cart is updated (-10).
I have the CTO saying we just need to "drop a few lines of javascript" onto the retailer online store. But I think this requires a Demandware plugin (especially updating a shopping cart), which involves signing up for a developer or partnership account (which takes some time), learning Demand Script and building the plugin, then releasing it to Saleforce's app store. Am I correct? Is this the only way to achieve this?
Just to be clear, the functionality is: the voucher code box needs to take a voucher number, an API request is made to our service to check what it's worth, then update the cart.
That's a pretty broad question but I'll try to explain.
First off, Demandware has gone all JavaScript.
But yeah, it has a huge API so the process sounds about right the way you've described it.
Ask your client whether they have a service integrator they're working with, because that would be their job. Your job would be to provide an API so that Demandware can check whether the voucher code is valid.
Demandware is a closed community so it's unlikely to learn it unless you're working in a company that is doing Demandware development.
Oh, and if it somehow ends up being your job to develop this, a tip from me - you probably need to get access to their site's files and use the LineItemCtnr class to createPriceAdjustment() :)
The first tip from me is that Demandware got bought by Salesforce and is now branded as Salesforce Commerce Cloud, so if you hear either, they are the same thing.
You are right that you are going to have to work with Salesforce to get going. They are a cloud provider and there is no self-hosted version of the platform, so in order to do any development, you need to have a relationship with them.
So then it depends on how you get access. If the client has any sandbox environments they are willing to share with you, you could do this as a custom build working in their environment, modifying cart code to make the service call as you ask and as the previous response suggests. But then (depending on the terms you have with the client) your company may or may not be able to reuse that code for other clients. You will also have to work with them to get that integrated into their existing site, get it deployed through their processes, etc.
If you want to write something that other Commerce Cloud clients could use, then you should set up your own partner relationship with Salesforce and get a software package they call a cartridge to what they call their LINK marketplace, which is basically a big clearinghouse of integrations like yours.
Technology-wise, the platform has gone through some changes recently, going from an XML-based GUI tool to specify business logic to using CommonJS scripts to define them. If your client is a new client within the last year or two they probably have this new fully javascript implementation, but if they are an older client than that they may be using the old way, which has a steeper (read: more proprietary) learning curve.
We look forward to maximise the use of Watson in our project. For this purpose we have to educate waston with all project documents, User manuals etc and use the Watson Conversation API possibly. But this seems a manual process of creating entities, intents and then dialogues. Is there a way Watson can identify potential entities and intents on its own? Or some way of automating or fast tracking the process of identifying entities, intents and creation of dialogues.
Can RPA (Robotic process automation) tools such as WinAuto or BluePrism be combined to use for this purpose ? Please suggest.
Thanks.
Depending on what you're seeking to do, Watson Conversation Service may not be the right answer for your use case.
What I read, looks more like a case for Watson Discovery Service, possibly combined with Watson Knowledge Studio.
You can read more about it here: https://www.ibm.com/blogs/watson/2016/12/watson-discovery-service-understand-data-scale-less-effort/?spMailingID=27478889&spUserID=MTY4Mzc0OTEyOTIxS0&spJobID=960800840&spReportId=OTYwODAwODQwS0
And secondly on the integration:
https://www.ibm.com/watson/developercloud/doc/wks/wks_mapublish.shtml
In the API documentation it states that methods are available to create all artefacts (workspaces, entities, intents, dialognodes etc..)
https://www.ibm.com/watson/developercloud/conversation/api/v1/?curl#update_workspace
But the SDK's don't support it yet (at least the Java SDK)
firstly I apologise if this is a ridiculously simple question to answer but it has been bothering me for a while.
I am trying to understand what salesforce actually is, I mean in technical terms. I have read the websites documentation and the wikipedia page but I am trying to understand what's behind all this fluffy terminology.
My understanding is that salesforce is a cloud based database which stores a very high volume of information and all salesforce apps consists of scripts that query this database and model them in different ways depending on the intended application, is this correct?
Thanks !
Software as a Service (SaaS)
To get program you need to download it, install, configure and so on. If your system have a lot of users it's very hard to configure ans support single user installation.
Imagine that you improved application, new release for example. You need update every instance.
With SaaS model you have a shared web application, that do the same thing as old downloadable one. But it's much easier to support it, because ideally there is just one instance of it.
Salesforce is a company that provides its own system by SaaS model, but not only. It is also a platform for developing new applications.
Requirements for archival type software
1. Data/Image/possibly video.... upload/search/retrevial/edit from web.
2. Easily implemented user defined Custom Fields
3. Easy backup.
4. Low cost ... either opensource or very low cost
I am a very novice programmer. My primary goal is to manage a collection and publish it to the web.
Options
A. Open source software such as collective access
Problems: Custom fields not supported. Continued support? Portablity of
database?
B. Use Microsoft Access and then use MVC or other development platforms to eventually
publish to the web.
Problems:Difficult to integrate to web?
C. Design my own MVC database application.
Problems:Difficult for novice programmer? Custom Fields and Upload of various data
formats difficult to implement?
Sounds like you are looking for a Digital Assets Management system. I found ResourceSpace (http://www.resourcespace.org/) and Razuna (http://www.razuna.org/) very useful for similar projects - both fall into your A category.
Requirements for archival type
software 1. Data/Image/possibly
video.... upload/search/retrevial/edit
from web. 2. Easily implemented user
defined Custom Fields 3. Easy backup.
4. Low cost ... either opensource or very low cost
Hi there,
As mentioned here before, but Razuna will satisfy your requirements quite well.
It can manage images, documents, videos and audios. It will share folderd and collections on the web with access permissions and will allow you to search among the different kind of assets as well.
Moreover, it can handle metadata of all this asset. It will not only read metadata, but also WRITE metadata, also. Furthermore, you can set the custom fields for each asset type and users will have a web interface to work with.
Razuna supports different databases (H2, MySQL, MS SQL and Oracle (soon DB2)) and let's you migrate from one db to another with ease (backup / restore option).
Best of it all: It is available under a open source license for you to deploy and enjoy today. You can get it at http://razuna.org.
Kind Regards,
Nitai
PS: I'm the main developer and founder of Razuna.
I'm hoping you can help.
I'm looking for a zero config multi-user datbase that my winforms application can easily upload to a webserver folder (together with 1 or 2 classic asp pages) and am looking for some suggestions/recommendations.
The idea is that the database will be used to collect feedback entered by people filling in the asp pages. The pages will write to the database using javascript.
The database will subsequently be downloaded again for processing once the responses are in.
In Summary:
It will mostly run in MS Windows environments.
I have a modest budget for this and do not mind paying for such a database.
No runtime licensing costs.
Should be xcopy - Once uploaded to a website folder it should be operational.
It should not have a dotnet CLR dependency.
It should support a resonable level of concurrent access. Average respondent count would be around 20-30 but one never knows.
Should be a reasonable size so that uploads/downloads to and from the site will be reasonably fast.
Would appreciate your suggestions/comments
Many thanks
Abz
To clarify - this is a desktop commercial application for feedback management in a vertical market. It uses SQL Server as the backing store.
The application currently provides feedback management from email and paper feedback. I now want to add web feedback capability. Getting users to to make their SQL servers accessible to a website is not at option at this time as I am want to make getting up and running as painless as possible.
I intend to release a web based implementation of the software in the near future but for now am looking at the above as a pragmatic way to provide web based feedback collection.
SQLite comes to mind. It meets all of your stated requirements, is open source, and has a liberal license (public domain).
http://sqlite.org/
I would use 'normal' database (say MySql, Postgresql, Firebird, etc.) on server. Instead of copying files to server your winforms application would create custom tables (or even custom databases). After collecting data you could just get it back to your application using plain old SQL.
why reinvent the wheel ? If you want to collect feedback and stuffs from users of your app and if they are connected to internet, it might be a better idea - and in the long term cheaper - to use a service like wufoo. We recently switched from homegrown setup to wufoo and are very pleased. Check it out.
Otherwise you might want to take a look at sqlite orfirebird. Both of them are very robust, and have ADO.NET providers. Firebird scales from a single user to full blown client server system and has no .NET dependency.
If you really don't want a DB/SQL Solution, you could try simple text files and ftp/xcopy files down and parse them into the back-office server as needed. ASP/VBScript or ASP.NET can create the files to store the basic feedback comments. Need to consider security of course!