Register
Login
Resources
Docs Blog Datasets Glossary Case Studies Tutorials & Webinars
Product
Data Engine LLMs Platform Enterprise
Pricing Explore
Connect to our Discord channel
Ezequiel Jadib 6b3df2c0b3
Update documentation links to point to new docs (#118)
6 years ago
..
3b8f26a2fb
Node - Samples ES5 compliant
7 years ago
3b8f26a2fb
Node - Samples ES5 compliant
7 years ago
6eeb67c4e4
Add repository, bugs and keywords to package.json
7 years ago
359c7c38a2
updated uuid package in node demo-search
7 years ago
74eb448cde
[Node] Update Emulator screenshots
7 years ago
7a204678e3
Node - Move environment vars for development to .env files
7 years ago
6b3df2c0b3
Update documentation links to point to new docs (#118)
6 years ago
7a204678e3
Node - Move environment vars for development to .env files
7 years ago
24e9d5a569
Apply fixes to samples after testing
7 years ago
359c7c38a2
updated uuid package in node demo-search
7 years ago

README.md

You have to be logged in to leave a comment. Sign In

Azure Search Samples

Depending on the nature of any given bot, sometimes you want to present structured dialogs that take users through tight paths, while in other cases you want to help users navigate a large amount of content. In many cases you'll have a mix of dialogs of both kinds in a single bot.

These samples illustrate how to approach dialogs that need to help the user navigate large amounts of content, creating a data-driven exploration experience.

Content is modeled as a catalog of items where each item has several attributes that may be used for navigation, keyword search or display.

Deploy to Azure

Prerequisites

The minimum prerequisites to run this sample are:

  • Latest Node.js with NPM. Download it from here.
  • The Bot Framework Emulator. To install the Bot Framework Emulator, download it from here. Please refer to this documentation article to know more about the Bot Framework Emulator.
  • [Recommended] Visual Studio Code for IntelliSense and debugging, download it from here for free.

The samples use Azure Search as the backend for these dialogs. This is an Azure service that offers most of the needed pieces of functionality, including keyword search, built-in linguistics, custom scoring, faceted navigation and more. Azure Search can also index content from various sources (Azure SQL DB, DocumentDB, Blob Storage, Table Storage), supports "push" indexing for other sources of data, and can crack open PDFs, Office documents and other formats containing unstructured data. The content catalog goes into an Azure Search index, which we can then query from dialogs.

As a good practice, all the Azure Search specific components are implemented in the SearchProviders/azure-search module. You can implement new providers be following the same API exposed by the module. See Implementing a new Search Provider below for more information.

Library & Dialogs

The samples include a bot library that contains a few different dialogs that are ready to use and can be customized as needed. The library is defined as the SearchDialogLibrary module and it includes two main dialogs that can be used directly:

  • Root dialog offers a complete keyword search + refine experience over a search index, and uses the other search dialogs as building blocks. Users can explore the catalog by refining (using facets) and by using keyword search. They can also select items and review their selection. At the end of this dialog a list of one or more selected items is returned.

  • Refine dialog helps users pick a refiner (facet). It's a simple wrapper around a "choice" prompt dialog to ensure you don't prompt users for a field you already refined on.

Samples

We included two samples here:

  1. RealEstateBot is a bot for exploring a real estate catalog.

It starts by taking an arbitrary set of keywords.

Emulator Facebook Skype
Search Search Search

From there you can go back and forth between keyword search and refinement using region, city and type of property.

Emulator Facebook Skype
Search Search Search

You can pick one or more properties and at the end you'll get a list of your choices (a real bot would probably contact your agent with your elections, or send you a summary email for future reference).

Emulator Facebook Skype
Search Search Search
  1. JobListingBot is a bot for browing a catalog of job offerings.

It starts by asking for a top-level refinement, a useful things to do in order to save users from an initial open-ended interation with the bot where they don't know what they can say.

Emulator Facebook Skype
Search Search Search

All samples target a shared, ready-to-use Azure Search service, so you don't need to provision your own to try these out.

Usage

In order to use the library you need to create an instance of it using the module's create() function and specify parameters that provides the search implementation and some level of customization for the search experience. Then, you register the returned library object with bot.use() and trigger the dialogs with SearchModule.begin(session) or SearchModule.refine(session, arguments).

var SearchLibrary = require('./SearchDialogLibrary');
var mySearchLibrary = SearchLibrary.create({
    multipleSelection: true,
    search: function (query) { return mySearchClient.search(query).then(mapResultSetFunction); },
    refiners: ['type', 'region']
});

bot.use(mySearchLibrary);

// Root dialog, triggers search and process its results
bot.dialog('/', [
    function (session) {
        // Triger search dialog
        SearchLibrary.begin(session);
    },
    function (session, args) {
        // Display selected results
        session.send(
            'Done! You selected these items: %s',
            args.selection.map(function (i) { return i.title; }).join(', '));
    }
]);

Parameters and Settings

SearchDialogLibrary.create(settings):Library

  • settings is required. The only detail really needed is the search function.
    • search is a required function. Accepts a query object and must return a Promise with a ISearchResults object (See Implementing a new Search Provider below).
    • multipleSelection (optional - default=true). Boolean value indicating if multiple selection should be allowed. When false, the dialog will end as soon as an item is selected.
    • pageSize (optional - default=5). Page size parameter passed to the search function.
    • refiners. (optional). Array of strings containing the facets/refiners allowed to use.
    • refineFormatter. (optional). Mapping function that must return a tuple for the refiner labels (to displayed to the user) and values (to send to the search provider).

You can also take a look at the TypeScript Type Definitions file.

Implementing a new Search Provider

This sample comes with an Azure Search provider ready to use. But there are cases were you'll want ot integrate another search engine or provider.

In order to integrate the Search Dialog library with a new search engine, you need to implement a function using the following signature:

function search: (query: Query) => PromiseLike<SearchResults>

Input query object with the following fields:

  • searchText: string

    The search string the user wants to look for.

  • pageSize: number

    Number of items to display in the results Carousel. If you are supporting Skype, the recommendation is 5 since it is Skype's limit for the Carousel component.

  • pageNumber: number

    The page number the use requested. You should use pageNumber and pageSize to calculate offset and limit values.

  • filters: Array of FacetFilter.

    List of filters/facets to apply on the search. An array of key/value objects. E.g.: [ { key: 'Region', value: 'US' }, { key: 'Type', value: 'House' } ]

  • facets: Array of string

    An array containing the facets you want to obtain possible values for. The search function should take this into account and proceed to search or retrieve sub-categories for the current search string.

The search must return a Promise, that once fulfilled, should resolve to a ISearchResults object. This object must have the following fields:

  • facets: Array of Facet. Each Facet object contains a key field with the name of the facet and an options field with an array of possible values. Each value is a tuple of key representing a possible facet value, and a count field with the total items for that facet.

  • results: Array of SearchHit. Each SearchHit is the object that will be displayed. The object may contain any number of fields, but these are required:

  • key: string. It is the Id of the object.

  • title: The name or title for the search result.

  • description: A description to be displayed about the item.

  • imageUrl: (optional) An url to be used along with the Hero Card when displaying the item.

You can take a look at a mocked search provider for reference.

More Information

To get more information about how to get started in Bot Builder for Node please review the following resources:

LimitationsThe functionality provided by the Bot Framework Activity can be used across many channels. Moreover, some special channel features can be unleashed using the Message.sourceEvent method.

The Bot Framework does its best to support the reuse of your Bot in as many channels as you want. However, due to the very nature of some of these channels, some features are not fully portable.

The features used in these samples are fully supported in the following channels:

  • Skype
  • Facebook
  • Microsoft Teams
  • Telegram
  • DirectLine
  • WebChat
  • Slack
  • GroupMe

They are also supported, with some limitations, in the following channel:

  • Email

On the other hand, they are not supported and the sample won't work as expected in the following channels:

  • SMS
  • Kik
Tip!

Press p or to see the previous file or, n or to see the next file

Comments

Loading...