DataSource PIM Feed User Guide

June 25, 2018

1. DataSource PIM Feed

1.1 Summary

A growing number of DataSource customers need to ingest DataSource’s structured content in enterprise Master Data Management or Product Information Management
solutions; those allowing for effective content ingestion/alignment and publishing of a single “reference” view of a product’s technical, marketing
or multimedia information.

The classic DataSource product data feed has limitations that made it challenging to import and manage in PIMs:

  • Not all product characteristics were delivered in their most granular/atomic form
  • DataSource Specifications being the result of a publishing process that involves aggregating multiple atomic attributes, they could not abide to PIM
    best-practices. Moreover, their underlying keys are not the same across languages and may change over time.
  • Adherence to international unit of measure standards
  • Attribute ordering

The DataSource PIM Feed provides all product technical characteristics in “granular atomic format”, i.e. the lowest granularity at which some characteristic
can be captured. This document describes the format of the DataSource PIM Feed option and compares it to the standard DataSource export.

1.2 PIM Feed vs. Standard DataSource Feed

1WorldSync Standard DataSource feed (aka Download.zip) differentiates between Searchable Attributes and Publishing Components:

  • Searchable Attributes: A subset of atomic attributes selected by 1WorldSync for each of the categories and delivered separately in atomic format. Searchable
    Attributes are intended to power parametric search, product comparison and custom categorization.
  • Publishing Components: The Standardized Description, Short Product Name, Main Specifications, and Extended Specifications contain ready-for-publishing
    product specifications built by aggregating specific atomic data elements. Most of the atomic technical attributes captured for a product are included
    in the Extended Specifications.

All Publishing Components are generated out of atomic technical attributes:

Example: Lenovo ThinkPad X1 Carbon 20A7 – Ultrabook – Core i7 4600U / 2.1 GHz – Windows 8.1 Pro 64-bit – 8 GB RAM – 180 GB SSD – no optical
drive – 14″ touchscreen 2560 x 1440 ( WQHD ) – Intel HD Graphics 4400

Structure: [Product Line] [Model] – [Notebook Type] – [Processor Type] [Processor Number] / [Clock Speed] – [Operating System] – [RAM
Size] RAM – [HDD Capacity] SSD – [Optical Drive] – [Diagonal Size] [Touchscreen] [Display Resolution] ( [Resolution Abbreviation] ) – [Graphics Processor]

DataSource with the PIM Feed option is upwardly compatible with the standard DataSource export.

The PIM feed as a variation of the DataSource feed that sees all attributes delivered as they were initially captured by 1WorldSync, and augmented with
additional model information that help answer common questions:

  • Is it an integer attribute?
  • Can it be have multiple values?
  • What is this unit’s relationship to existing standards?
  • In which order should I display a product’s attributes?
  • How can I compose higher level attributes from granular attributes?

The PIM feed also gives access to new attributes that DataSource could not deliver in the past. For example, a TV’s connectivity information used to be
shipped in the specifications (3 x USB 2.0) but not entirely as attributes. This is now possible with the definition of new “secondary” attributes,
in this case “USB Ports Qty”.

The table below highlights the main differences between the Standard and PIM feeds:

PIM Feed Standard DataSource Feed
Publishing Components DataSource customers might choose to use 1WorldSync generated product specifications or create specification display rules on their own.

DataSource feed contains generated technical specifications including:

  • Standardized Description
  • Short Product Name
  • Main Specifications
  • Extended Specifications
Searchable Attributes All product attributes are delivered in atomic/granular format.

Only a small subset (up to 25%) of attributes is delivered in atomic/granular format.

The rest of the attributes are delivered as a part of technical specifications.
Data Model Maintenance

High data model maintenance cost for the customer as each DataSource attribute needs to be associated with an attribute in customer’s
taxonomy.

PIM feed contains full model meta-data and attribute life-cycle information.

Low data model maintenance cost for the customer: only key attributes need to be mapped (mainly for search facets and categorization).

DataSource
specifications are intended to be used as indivisible ready-for publishing data components.

1.3 Downsides of PIM Feed

DataSource Product Data Model is constantly evolving. 1WorldSync team keeps adding new attributes and values, deactivates old/obsolete PDM elements, replaces
value lists, etc.

There is a relatively small subset of Managed attributes (mainly searchable ones) that get changed only after advance notice. But any of unmanaged attributes
can be changed at any point of time.

Depending on the level and complexity of integration and the ecommerce website, any DataSource customer using PIM Feed (or old Legacy Atomic Feed) need
to trace all PDM changes to ensure they change their integration rules accordingly. Otherwise some of the custom categorization, product search or
attribute display rules might deteriorate with time.

Bottom line: PIM Feed provides certain extra flexibility around data structure and format customization, but at the same time requires a lot of extra constant
monitoring and maintenance by those DataSource customers that choose to take advantage of it.

1.4 PIM Feed vs. Legacy Atomic Feed

PIM Feed Legacy Atomic Feed
Attribute
Display 		PIM Feeds ordering for Attribute Groups and Attributes within each group. This order can be used for technical specifications display,
side by side product comparison and parametric search. 		Atomic feed does not contain attribute order, which makes it difficult for DataSource customers to figure the logical display order for
the attributes in product specifications.
Attribute
Groups 		Attribute names and attribute group names are provided separately. Attribute names include attribute group names, which requires additional logic on reseller side to cleanse and simplify them.
Composable
Attributes 		PIM Feed contains both Composable attributes, and Composed attributes derived from them. Therefore, it is not required to integrate Composable
attributes any longer. 		Most of the PIM systems cannot ingest Composable attributes (repeating attribute sets with SetNo field).
Special
Attributes 		PIM feed clearly identifies the attributes to be Displayed in product specifications. Some service attributes are not intended to be displayed in product specs. They might look weird for the end users if included in the specs.
Duplicated
Attributes 		PIM feed clearly identifies the attributes to be Displayed in product specifications. Some characteristics are duplicated in two or three different attributes intended for different purposes: parametric search, product title
or specs.
Duplicated
Attributes 		PIM Feed clearly differentiates deactivated attributes. Some attributes in the feed are deactivated. Atomic Feed does not differentiate active and deactivated/legacy attributes.

1.5 Tables

cds_PIM_atr Main table with all product attributes in atomic format for each SKU (product) in the data feed.
cds_PIM_atr_composable Additional table containing just Composable attributes with set number (SetNo) field.
cds_PIM_atr_voc Vocabulary table with all attribute group names, attribute names, values and units.
cds_PIM_atr_standard_units Table with conversion rules from DataSource units to the units form various international standards.
cds_PIM_atr_order Meta-data table with attribute order and attribute groups for each category.
cds_PIM_atr_model Meta-data table with data model structure: attribute types, attribute grouping and value list format.

1.6 Data Model Elements

Category E.g. “Notebooks”
Attribute Group E.g. “Processor”
Attribute E.g. “Clock Speed”
Value E.g. “2.7”
Unit E.g. “GHz”
Unit List E.g. “Frequency Units”
NNV Value NNV value is generated for numeric (integer or float) attributes. If an attribute does not have a unit list, NNV is the same as the attribute
value, only without thousand comma separators. If an attribute has a unit list, NNV value is expressed in the lowest unit denominator.

E.g. “2700000000” expressed in Hz base unit for 2.7 GHz.

1.7 Attribute Types

Single-value The Attribute can have only one value for any given SKU.
Multi-value The Attribute can have multiple values for the same SKU.
Composable The Attribute belongs to Composable Attribute Group that is repeated for the same SKU multiple times. For example, Interfaces Attribute
Group has multiple attributes: Type (USB 3.0), Qty (3), Location (in front). Then entire attribute set needs to be repeated once again
for Headphones port, HDMI port, etc.
Composed Composed attributes are text multi-value attributes generated from Composable attributes.

1.8 Attribute Format

Text Text attributes always have controlled standardized value lists. The length of the values is limited by 255 characters.
Integer Some of the integer attributes have comma thousand separators for display purposes.

E.g. “5000”, “14000”, “61,380”, “921,600”.

Float Float attributes use period as decimal separator. Some of the float attributes have comma thousand separators.

E.g. “44.1”, “28.8”, “2,646.00”, “2,000.00”.
Boolean Boolean attributes typically have “yes” or “no” values (consistently translated to supported languages). Some of the Boolean attributes
have only one value “yes” and are left empty if the feature does not exist or the attribute is not applicable.

Besides that, there are few service Boolean attributes with “true” and “false” values.

2. Export Schema

3. Feed Format

3.1 Feed Encoding

All tables are delivered as UTF-8, UTF16 and Western European Windows code pages.

If a customer is subscribed to Windows encoding, Asian and Eastern European translations will not be included in the vocabulary file.

3.2 cds_PIM_atr

Column Type DB Description
ProdID Varchar(40) PK Product ID
AtrID Varchar(10) PK Attribute ID
ValID Varchar(10) PK Value ID
UnitID Varchar(10) NULL Unit ID
NNV Float NULL Normalized numerical value, expressed in the attribute’s base unit

This table is incremental synchronized with cds_Prod increment.

3.3 cds_PIM_atr_composable

Column Type DB Description
ProdID Varchar(40) PK Product ID
AtrID Varchar(10) PK Attribute ID of Composable Attributes
ValID Varchar(10) PK Value ID
UnitID Varchar(10) NULL Unit ID
NNV Float NULL Normalized numerical value, expressed in the attribute’s base unit
SetNo Integer PK Number of attribute set in a Composable Attribute Group

This table is incremental synchronized with cds_Prod increment.

3.4 cds_PIM_atr_voc

Column Type DB Description
ID Varchar(10) PK ID of attribute group, attribute, value, unit list or unit
LanguageCode Varchar(20) PK Same codes as we use in digital content and ACC. This field refers to the languages defined in Cds_Languages table.
Text Nvarchar(2000) Textual value for attribute name, value, unit-group name or unit name
ShortText Nvarchar(255) NULL Short value when it exists

This table is incremental synchronized with cds_Prod increment.

3.5 cds_PIM_atr_standard_units

This table gives mappings between DataSource units and standards for Units of Measure (ISO, unece). Note that there might be several DataSource Unit IDs
mapped to a single standard record as 1WorldSync creates units based on a capture context – for example, length units for information captured on an
appliance (fridge) will belong to different unit lists than those captured for a cable or laptop.

Column Type DB Description
UnitID Varchar(10) PK DataSource Unit ID
UnitListID Varchar(10) DataSource Unit List ID
BaseUnitID Varchar(10) Base DataSource Unit ID
StandardTitle Varchar(200) PK Title of a standard or a system of units
StandardUnitName Varchar(100) NULL Standard unit name
StandartUnitCode Varchar(50) NULL Standard unit code
Description Varchar(2000) NULL English description of the unit
ConversionMultiplier Float Conversion factor between the unit and its corresponding base-unit
ConversionIsDivision Bit Conversion requires dividing ConversionMultiplier by the attribute value. E.g. conversion of car fuel consumption between l/100km and mpg.
ConversionAddition Float NULL Constant that needs to be added to converted value. E.g. conversion from F to C.

This table is non-incremental: always delivered in full.

3.6 cds_PIM_atr_order

Column Type DB Description
CattID Char(2) PK DataSource 2-letter category code (classic categories)
AtrID Varchar(10) PK Attribute ID
GroupID Varchar(10) Attribute Group ID
GroupOrder Integer Display order for attribute group
AtrOrder Integer Display order for attribute within group
DisplayInTitle Bit When 1, attribute is recommended to be displayed in specifications. Use 0 when it is not. This field is based on inclusion of the attribute
in Standardized Description.
DisplayInMainSpecs Bit When 1, attribute is recommended to be displayed in specifications. Use 0 when it is not.
DisplayInExtendedSpecs Bit When 1, attribute is recommended to be displayed in specifications. Use 0 when it is not.
IsManaged Bit When 1, this attribute will be part of model change communications when introduced or deactivated. An attribute is managed when it is searchable.
IsDisabled Bit When 1, attribute has been disabled and may not ship in the future. 0 for active attributes.

This table is non-incremental: always delivered in full.

3.7 cds_PIM_atr_model

Column Type DB Description
AtrID Varchar(10) PK Attribute ID
GroupID Varchar(10) Attribute Group ID
AttributeType Varchar(50) Attribute type: single-value or multi-value. This field is generated based on attribute type. All Composable attributes are single-value.
All Composed attributes are multi-value.
ValueType Integer Attribute value format (integer, float, text, Boolean)
IsComposable Bit Composable attributes (aka repeating attributes) are the attributes from repeating attribute groups used to generate Composed attributes.
IsComposed Bit Composed attributes are fake attributes generated based on Composable Attribute Group display rules in Extended Specifications, when entire
attribute group is presented as one multi-value attribute.
IsLocalized Bit Is this attribute localized: if 0 then all values for this attribute translate the same way. If at least one of attribute values or units
is translatable, then the attribute is localized.

This table is non-incremental: always delivered in full.