Find: Search API

lang

placement

query

The text to search.  

Example: query=shoes

There is a limit to the size of a query.  It is currently limited to 75 characters. 

rows

start

The starting position of the results you want for the specified query.  It's a zero indexed system.  

For example if you want the first 20 results, you will set start=0, if you want the next 20 results (21-40), you will set start=20.

callback

Name of the JavaScript function that JSON data will be passed to. It must be specified for JSONP. The value of this parameter is used as the name of the js function in the response.

Example: callback=products_returned

channelId

 Channel ID is used by some Algonomy reports to differentiate channels such as mobile or desktop.  Example values are: WEB, WEB_PHONE, WEB_TABLET.  It can be any API client key that an Algonomy consultant has set up for a customer.  

facet

Specify which fields/attributes you want facets for in the response.  If you don't specify any, then all attributes configured to be facetable (in the dashboard) will be returned in the response. 

For a list of all applicable fields see section below called "Fields".

Controls the number of category levels returned in the Category Facet filter.

facetDepth < 0 - Depth parameter is not considered, Standard behavior is applied, i.e. only the top level is returned on the first request and the children are only returned after the top level selection.

facetDepth == 0 - Return no categories to filter.

facetDepth > 0 - The category depth is built until max depth value level or the depth specified in the parameter is reached. Example: facetDepth=5 will return 5 depth level if the category hierarchy has all that levels. If the hierarchy has only 3 levels, then only 3 depth level will be returned.

Default value = 100
Can be a +/- integer value
 
A negative value means that Solr will return unlimited number of constraint counts.
 
More information :
https://lucene.apache.org/solr/guide/6_6/faceting.html#Faceting-Thefacet.limitParameter

facet.limit is only applied to the following responseStyles:  fullResponse and facetsOnly.  

NOTE: facetDepth has priority over facet.limit and facet.mincount. When facetDepth is present, facet.limit and facet.mincount are ignored for category field (product_category_name)

Default value = 0
Can be an integer value.
 
More information :
 
https://lucene.apache.org/solr/guide/6_6/faceting.html#Faceting-Thefacet.mincountParameter

NOTE: facetDepth has priority over facet.limit and facet.mincount. When facetDepth is present, facet.limit and facet.mincount are ignored for category field (product_category_name)

filter

Apply a filter to the search term.   You can apply multiple filters.  For a list of all applicable fields see section below called "Fields".  The response (see section called "Response" below) also provides you a list of facets and how to filter by the facets.

Here's an example of how you filter for the brands "adidas" and "puma"

filter=product_brand:"adidas"&filter=product_brand:"puma"

Another filter example with AND and OR:

filter={!tag=material q.op=OR}material:("Leather" OR "Cotton").

The "tag"  filter syntax is used return all the different options for a facet even after filtering. 

Specifies the call type from the client. Default is 'direct'.

direct - If log is true, record the search request event within the analytics system and also forward the search request event to the real time statistics service.

overlay - If log is true, record the search request event within the analytics system, but do not forward the search request event to the real time statistics service. In that case, the searchTrackingUrl should be used to provide real time statistics.

Field List

The field list parameter limits the information included in a result set to a specified list of fields. The result set will only include the fields specified in the field list plus id and score (which are always returned). Each field must be "storable" or its values won't be returned.  The default is to return all storable fields.

The field list is specified as a comma-separated list of field names. 

Example: fl=size,color      will return:  id  score  size color

If blank, wildcard or null are used, the following will occur:

fl=   (blank)  Returns all storable fields

fl=*   Returns all storable fields

fl=null    No fields are returned

Note:

To return clickURL, please use linkID.  Using linkId will return both the clickURL and linkID.

clickURL is a derived attribute and doesn't have search attributes associated with it such as "storable". 

Example:

Example: fl=size,color,linkId      will return:  id  score  size color linkID  clickURL

log

If set to true this will log the search request event for analytics.  The default behavior is for this to be false.

NOTE: Only the values "true" and "false" will work.

N-grams are like a sliding window that moves across a word - a continuous sequence of characters of a specified length which in Find is defaulted to 1.  Ngram search is often used with an auto-complete/type ahead feature. For example, as a shopper types each character in the search box, there is a match based on the characters entered.  Some Find customers use the Find API for auto-complete of products instead of the autocomplete API.  It is important for them to be able to set ngramSearch=true as part of the Find query.   It is recommended to set to ngram=false for general product queries. 

There can be a system wide setting for ngramSearch in the site configuration which an Algonomy consultant sets up.  If it is not set up in the site configuration, the default is true. 

This parameter will over-ride the setting in the site configuration as well as the default value.

ngramSearch=true/false. 

Shopper’s referrer prior to viewing the page. Used for reporting and merchandising. This is highly recommended.

Example: pref=http://www.google.com

region

The responseStyle parameter controls the result set that is returned.  By default all fields which are tagged as "storable" are returned as well as facets (facetable).  Personalization and boosts will also be applied if they are setup and turned on.  The responseStyle parameter has a number of options which are detailed below.

fullResponse - Default. Result set limit is 512 rows.

idAndScores  - Returns just the product Ids and scores. The result set limit is up to 10,000.

facetsOnly - Returns just facets and not product or content.  There is also no limit on facets being returned. However the default number of items returned within a facet is 100 and can be controlled by the facet.limit parameter.   Since products and content are not returned, the rows parameter is automatically set to 0.  There is no need to provide the rows parameter in a facetsOnly query.  

productsOnly - Returns products only. Does not return facets and also does not apply personalization, boosts, links, merchandising rules or spell check.  One use case is when personalization is not required and the customer wants to display "new" products and can sort by date.   The purpose is to only bring back the required information and to avoid unnecessary processing.

Example of idAndScores: 

    Request: https://...start=0&rows=10000&facet=0&query=phone&responseStyle=idAndScores

Response:

....{

     "docs": [

        {   "score": 1,

           "id": "58481" }

......}

Identifies a single visit by a shopper.  Sessions are used by behavioral models (to scope user behavior in a shopping session.) and reporting metrics.

Example: sessionId=93484

sort

Specify how you want to sort a result list of either products or reference content.  The default is a personalized sort based on "score" (sort=score desc). You can instead sort by any attribute that you specified in the search attributes to be sortable.  For a list of all applicable fields go to the "search attributes configuration" in the portal. 

We recommend that in addition to the personalized relevant sort (the default) you give options for shoppers to sort by these fields

• product_rating (Average rating from reviews)
• product_effectiveprice_cents (Combined price and sale price)
• product_release_date (The newness of the product)

Also,  specify whether you want the sort to be in ascending or descending order.  Allowed values are: "asc" or "desc".  One must be specified.

Example: sort=product_release_date desc

You can specify more than one field, and the field name can have spaces. You can also still have the score as the primary sort followed by a secondary sort as in this example:

sort=score desc, field name asc

Note: Multi-value attributes are not sortable.  

If the sort parameter is not correct the default will be used which is "score desc". 

Find will spellcheck a query term when the the initial query has 0 results.   It then executes a second query using the recommended spellchecked term. The spellcheck is based on all indexed searchable attributes except for wisdom of crowd attributes and product_external_id. 

The default is "jaro" which is a similarity algorithm to compare short strings. 

The following spellcheck options are provided in order to return the best suggested term(s) and meet customer expectations. 

     jaro - default, JaroWinklerDistance is a similarity algorithm to compare short strings. 

     off - spellcheck is turned off, useful when the query is a bar code or other product code.  

     hits - return the suggested term that has the most hits

     suggest - a list of suggestions is returned and is useful when implementing "did you mean". 

Below are further details and examples.

jaro   - default - not explicitly provided in query

If the spellcheck parameter is not explicitly set, the "spellchecked" field is returned.

query=seattle1
Response will include: 

"spellchecked" : "seattle"

jaro - parameter provided in query - equivalent result to "query=seattle1".

Note that the "spellcheck" parameter will provide the type of spellcheck and the suggested term. 

query=seattle1&spellcheck=jaro
Response will include:

"spellchecked" : "seattle"
"spellcheck" : {
  "spellcheckType" : "jaro"
  "spellchecked" : "seattle"
}

off

query=seattle1&spellcheck=off
Response -  "docs" is empty

"spellcheck" : {
  "spellcheckType" : "off"
}

hits  - see the "suggest" list below to understand why "settler" is returned

query=seattle1&spellcheck=hits
Response will include: 

"spellchecked" : "settler"
"spellcheck" : {
  "spellcheckType" : "hits"
  "spellchecked" : "settler"
}

suggest

A set of suggested terms are returned that could be used for "did you mean" and to build an alternate query. This option does not automatically execute a second query as is the case for "jaro" and "hits".   Two lists are returned:  "spellcheckSuggestions" and  "suggestionsWithHits"  (number of hits  for suggested terms is provided).  

query=seattle1&spellcheck=suggest
Response - "docs" is empty 

"spellcheck" : {
  "spellcheckSuggestions" : {
    "seattle1" :  ["seattle", "seat tl e1", "settler"]
  },
  "suggestionsWithHits": {"seattle":8,  "settler":20},
  "spellcheckType": "suggest"
}

suggest - multiple search terms

query=seattle1 washington1&spellcheck=suggest
Response -  "docs" is empty

"spellcheck" : {
  "spellcheckSuggestions" : {
    "seattle1" :  ["seattle", "seat tl e1", "settler"]
    "washington1" :  ["washington", "washing to n1"]
  },
  "suggestionsWithHits": {"seattle washington":8},
  "spellcheckType": "suggest"
}

By default the score for each item in a result set is based on the "max" score from one field (plus personalization, boosts etc also play a role).  Often many results will have the same score.  "Tie" is a Solr parameter which takes into account the hits on other fields.  It applies the following calculation:

score = [score of the top scoring subquery] + tie * (sum of other hitting subqueries)

It is recommended to set  "tie" to a low value 0.1 in order that the sum of the other fields doesn't over power the key field with the most hits.  Values can range from 0.0 to 1.0 with the default set to 0.0

"tie" can be a parameter in a Find API query or in the search configuration in the portal.  If in the Find API, it will over-ride the value in the search configuration portal. 

Example Query: ...query=shoes&rows=15&tie=0.1

Further details: https://lucene.apache.org/solr/guide/6_6/the-dismax-query-parser.html#TheDisMaxQueryParser-Thetie_TieBreaker_Parameter

User ID. A unique string to identify each shopper (user). All shopper behavior is stored using this key. It is case-sensitive, and should be the same user ID sent to RichRelevance in other applications.

Example: userId=0982347

If no userId is given, search results will be based on session based history (sessionId) and previous view/purchase behavior through the rcs parameter/cookie implementation.  The search results will still take Wisdom of the Crowd behavior into account.

In almost all cases there should only be 1 placement requested for search.  This object will contain all the information needed to render the search results on a page/app.

This is a nested field that contains multiple fields:

1. searchTrackingUrl:  A URL used for real time search request statistics within the Overlay client-side application.  This URL should be invoked by the Overlay application. At the same time, the Overlay application issues search requests with findCallType = 'overlay'.
2. docs:  The list of products that match the search term - 

   NOTE: if the clickUrl is used directly on the product and requires the redirect to the product page  the parameter &redirect=true needs to be added to each clickUrl

3. numFound: How many products in the catalog match the search term and any filters applied.  This is what is used to communicate to the end user how many products match their search term.
4. placement:  The name of this placement.
5. spellchecked: If there was a typo that was auto-corrected by our engine, the corrected term will be specified here.  If no spell correction was needed this value will be empty or 'null'.
6. facets:  List of all facets that are applicable for the products that match this search term
7. addtoCartParams: parameters that should be appended to add to cart event firing from product detail page of the search result product. 
8. links: configured links (URLs) with the search term that was specified the request
   1. banner: link to the banner image that should be displayed along with the search results.
   2. directlink: link to another page where shopper should be asked to follow.
   3. sponsored: link to sponsored content that should be displayed 

facets are returned as part of the placements structure. They power the faceted search navigation experience. 

Facets contains these fields:

1. values : array of possible facet options for a give type

2. filter : filter query that should be used when the shopper selects this facet option. filter query should be passed as it is to the filter parameter of the subsequent search request

3. count : Number of items matching this facet option. In other words, number of items to expect when filter for this facet is applied to the subsequent search request.

4. value : Display value for the facet

5. child : For hierarchical attributes like categories, Find provides a way to let shoppers navigate through the hierarchy by providing immediate next child categories in the facet values. This is invoked by providing higher level category as a filter. For example, filter=categoryId:root will provide 1st level children of root as facets in the search response.

6. Price min - max : For price facets, Find returns min and max price values for the current search results, to be able to power a price selection slider kind of UI control instead of option boxes for each individual price ranges.