bintime_test 00000-895612 Product information

Category
Night creams
Type
Product information
Icecat Search Engine (ISE)
1. SEARCH
1.1 Summary
Within new ISE implementation, there will be one common search template for category, search
results and vendor page with slight differences in logic and look&feel.
Search/Vendor/Category results pages will contain:
List of results
Filters
Comparison
Pagening
Sortings
Vendor and Category results pages will also contain:
Vendor/Category logo
Vendor/Category description
Search input field should be present on the following pages:
- home page
- category page
- sub-category page
- product page
- user Interface
- privacy
- forum page*
Forum search will be developed in the future and requires separate documentation and detailed
explanation. This document includes description of content search engine only.
1.2 Data indexing
Search index of ISE will consist of the following content:
MPN - numeric, alphabetic or alphanumeric identifier of manufacturer's product. Note!
Mappings of part codes must be also indexed (can be added to thesaurus).
e.g. BDT4651TO/00
Brand - alphabetic, alphanumeric or rarely just numerical value - name of brand. Note!
Mappings of brands must be also indexed (can be added to thesaurus).
e.g. Philips
Category / virtual category - alphabetic name that defines Icecat category or virtual category
name. Note! All category names should be added to index (eg notebook andd notebooks)
(Multilingual!)
Model name - alphabetic, alphanumeric or just numerical value that defines the model name of
product
EAN/UPC code - numerical value that contains from 8, 12 or 13 numbers. Unique for one
specific product. So one EAN/UPC code can not relate to more than 1 data-sheet. Note!
Alternatives of EAN codes must be also indexed (can be added to thesaurus)
Example: 8712581618667
Product title - alphabetical or alphanumerical value that contains from Brand name, Family
name (if present), Model name and sometimes category name (if model name equals to
manufacturer part code). As far as Product title is repeating the parts of it which are already in
index, test with indexing product title and without and choose the best option.
Example: ASUS A55 A55VD-SX521H notebook where: ASUS - brand name / A55 - family name
/ A55VD-SX521H - family name / notebook - category (present because model name = man)
Short description. Multilingual
The above properties can be divided between two indexes:
index, independed from languages/locales (MPN, Manufacturer, EAN/UPC etc)
index, depended on languages/locales (Category, Family series, Short description,
Model name). This index will contain all possible lnaguage translations, available per
product.
There are 2 indexes: Full (TTL about 1 hour) and Delta (TTL about 15 mins)
During the procedure of adding data to index, we can influence the rating of the products by the
following formula:
Product rating = ( K1*Rating(Category)+K2*Rating(Vendor)+K3*Rating(Product) ) / Denominator
Where koeficients K1,K2 and K3 can be set in config.
Denominator - normalization of received amounts
Rating of Categories and Vendors can be set manually or automatically basedon some
calculatable values (sponsor/not sponsor, total amount of page views of all products per
category etc)
1. 3 Data selection from index
Every search string is divided by spaces and each word is being searched in index. After this,
Lucene returns the list of suitable (relevant) products, and products data is taken from MySQL
DB.
During search, selection of data from Index can be done in 3 ways:
By Name
Example: “asus notebook”
((attr_mpn:asus*^10 || attr_sku:asus*^10|| attr_name:asus*^5 || attr_description:asus*^2||
attr_supplier_name:asus*^5|| attr_supplier_description:asus*^2 || attr_manufacturer:asus*^10 ||
attr_cat_name:asus*^10 || attr_full_field:asus*)
AND
(attr_mpn:notebook*^10 || attr_sku:notebook*^10 || attr_name:notebook*^5 ||
attr_description:notebook*^5 || attr_supplier_name:notebook*^10 ||
attr_supplier_description:notebook* || attr_manufacturer:notebook*^10 ||
attr_cat_name:notebook*^10 || attr_full_field:notebook*))
Search provides the relevance level of matching documents based on the terms found.
^X is boost factor (a number) at the end of the term - the higher the boost factor, the more
relevant the term will be. Boosting allows to control the relevance of a document by boosting its
term.
The boost factors can be adjusted easily in config.
By this type of search the Search page is dispayed
By Brand (perfect match by brand name)
(manufacturer_s:"asus")
By this type of search the Maufacturer page is dispayed
By Category (perfect match by category name)
(cat:"43171801")
By this type of search the Category page is dispayed
1.4 Search by content type
Visually:
Possibility to search by content, image and video can be shown in several ways: as a dropdown
behind search (content type by default) as on search.bintime.com
or
Filters above search results (content type by default) as on google.com
Technically:
There should be a possibility foreseen to search within different types of content:
- Default search by content. This is a default search which involves searching in MPN, Brand,
Category, Model name, EAN/UPC, product title
- Search by image. As far as Icecat has no image-specific meta data, the search per image will
be in reality the same search as search by content but with different visualisation of the search
results. It is possible to show only main image or all images from the gallery of products from
search results.
- Search by video. As far as there is video description available in Icecat DB, this field can be
added to video search index.
1.5 Search by exact match of Category/Vendor
If user searches for the keyword which has the same spelling as category name or vendor
name, he/she is directed to the vendor page or category page of the shop. When user can
continue searching by filters refining.
1.6 Search by MPN:
Search query using MPN may return more than one result. For instance, search by CB323EE
part code gives several results:
CB 323 EE
CB 323 EE # ABB
CB 323 EE #301
Important: search engine must respect “mapping part codes”, so m_prod_id’s
If visitor search by a part of manufacturer product number, for example CB323 search engine
should return all possible
1.7 Fuzzy search:
In case when user searches for some keyword, which is similar to MPN or EAN in search index,
this similar keyword should be suggested for the user.
Example: If search request is CB323EE#ABC or CB323EEABC so doesn’t fully correspond to
manufacturer part numbers in Icecat database, system should provide user with suggestion as
google does - “You are looking for CB323EE#ABC, maybe you mean CB323EE”
Important notes:
Single and plural forms detection. Both single and plural forms of the nouns are
recognised by the search and are operated in the same way.
If search request doesn’t return anything, system should display tokenized (so
connected with dictionary) text: “Please try to refine your search query”.
Inactive products should be hidden under special link, some kind of “Show more” link in
the bottom of search results.
Undescribed (NOEDITOR group) products should be hidden by default, but only a
perfect hit (search by MPN or GTIN (EAN/UPC) by a customer may return NOEDITOR
products.
Additional optional features:
Search by parts. This features allows to split long search keywords which lead to zero results
into several shorter search phrases which have the results. There is amount of results indication
next to each part of the long keyword.
Related searches. When user is directed to the search results page, he/she is also suggested to
try related searches, which are based on automated suggestions.
2. AUTOSUGGESTIONS
2.1 Autosuggestions
Visually:
Autosuggestions can consist of phrasal list suggestions and product list block.
Technically:
The suggestions below the search field appear as soon as the user starts typing the search
keyword.
Autosuggestions are manageable from BO, here are the features which can be changed by
admin from BO:
choose between phrases and products list views or have both views combined
change prio for specific suggestions
add/edit/delete suggestions
enable/disable suggestions
select Brand or Category grouping in the product list block of suggestions
Phrasal autosuggestions – phrasal suggestions help users to define their search, basing on
relevancy and popularity of categories and vendors. The most appropriate autosuggestion is
also displayed in 2 top categories.
Search as you type by products – products list, grouped by vendor or category gives the
opportunity to jump from search directly onto the product page of the desired product.
Logic of autosuggestions creation:
Suggestions are created based on product title, categories and vendor names. There can be
any existing rating used for better relevance of the search (product rating, category or vendor
rating – both manual or automatically calculated)
Product names: different combinations of words in product name
Example:
Product name “word1 word2 word3 word4”
Suggestions:
word1 word2
word1 word3
word1 word4
word1 word2 word3
word2 word2 word4
Rules of selection for name-based suggestions:
1) To keep only relevant suggestions in database, the procedure deletes all phrases that do not
meet next criteria:
Less then 8 matches in db for 1 word phrase.
Less then 6 matches in db for 2 word phrase.
Less then 4 matches in db for 3 word phrase.
(the above parameters are configurable and can be changed during tests on live data)
2) Using quantile (points taken at regular intervals from the cumulative distribution function)
Values behind 20% are dropping out.
Name-based suggestions receive Type 0 flag
Vendors and Categories:
Next to names, suggestions are built based on vendors and categories combinations.
Vendors, Categories → Type 2
Vendors + Categories, Categories + Vendors → Type 1
Sorting of suggestions:
High level Sorting is based on types
Type 2 (limited to 2 suggestions). Inner sorting: amount of words (asc), rating (desc), results
amount (desc), alphabetic(asc)
Tape 1 (limited to 2 suggestions)
Type 0 (limited are configured in BO)
2.2 Thesaurus
There must be a thesaurus created with the rules for synonyms identifications. The rules will be
added manually (or automatically in case of some common logic is outligned J).
For instance, in order to have a good hint for “Galaxy S III”, the following rule should be created
III = iii, 3, Three (symmetrical rule)
Thesaurus will be used for autosuggestions creation.
2.3 Spell check
Smart function of misspell detection allows to autocorrect wrongly spelled words and provide the
possible correct options in suggestions. Configurable accuracy value is used by the spell
checking implementation to decide whether a result is worthwhile or not. The accuracy value
can be changed during tests.
3. SORTING
3.1 Default sorting
MPN:
In this case user should be provided with the most relevant results of his inquiry, so display
order should be the following:
1) highest-relevant value => CB323EE
2) value with highest product view index (or lowest price / or highest number of offers)
Brand (exact match with a single brand):
Search query based on brand name should return the list of product data-sheets of respective
supplier. In this case it’s necessary to show the most popular products in scope of one brand.
Popularity should be based on three parameters:
1. product activity
2. product quality - ICECAT is prioritized as more described
3. number of product displays (as one of ways to define popularity)
3.2 Other sortings
1) Price (asc, desc)
2) Popularity (Number of products displays)
3) Market reach (Number of offers from merchants)
By other sorting product activity and quality is used as inner sorting.
4. FILTERS
4.1 Filters display
In order to help users to refine search results, filters will be displayed on
Search/Vendor/Category results pages.
Search result filters:
Brand
Category (1
st
level category by default. When selecting top category, its subcategories
are displayed and so on)
Price (autogenerated pricebreaks or min and max input values)
Features list (current list of features per subcategory). Note! This list of features will
appear only if 2
nd
, 3
rd
etc level
category is selected (in category dropdon) AND if category has
products AND if category has searchable features
Example: http://search.bintime.com/search/category/43171801
(admin : skdf$#&&%tg)
Top Category filters:
Brand
Categories of 2
nd
level
Price (autogenerated pricebreaks or min and max input values)
Features list (current list of features per subcategory). Note! This list of features will
appear only if category is selected (in category dropdon) AND if category has products AND if
category has searchable features
If any filter is selected, it would be good to have Category description/logo collapsed
nicely.
Example: http://search.bintime.com/search/category/43201513
(admin : skdf$#&&%tg)
Sub Category filters:
Brand
Categories of 2
nd
level
Price (autogenerated pricebreaks or min and max input values)
Features list (current list of features per subcategory). Note! This list of features will
appear only if category is selected (in category dropdon) AND if category has products AND if
category has searchable features
Example: http://search.bintime.com/search/category/43201513
(admin : skdf$#&&%tg)
Vendor p a g e filters:
Category (1
st
level category by default. When selecting top category, its subcategories
are displayed and so on)
Price (autogenerated pricebreaks or min and max input values)
Features list (current list of features per subcategory). Note! This list of features will
appear only if category is selected (in category dropdon) AND if category has products AND if
category has searchable features
If any filter is selected, it would be good to have Vendor description/logo collapsed
nicely.
Example: http://search.bintime.com/search/?category=&yt0=&keyword=acer
(admin : skdf$#&&%tg)
4.2 Rules for filters
Selected filters
Selected values of filters are displayed on the top of left column of filters with a possibility to
deselect all selections.
http://rozetka.com.ua/notebooks/c80004/filter/preset=netbooks;producer=acer;25800=20879/
Multiselect on values
Most of the features can have multi-value selection (e.g. to shw notebooks with 13” and 15”, so
2 values selected). Some filters have one-value selection (e.g. price breaks, categories).
Page reload by values selection
Search result page should not be reloaded fully after filters selection. Only the area with
products and filters should be reloaded.
Long list of values
Limit list of values to be displayed per feature to 5, show others under “show more” link.
Show products per page
Show 20 products per page. Add possibility to show 10, 20, 40, 100 products per page. Keep
this selection when changing filters, using pagening and other actions.
Amount of products per value
There can be a counter next to each value, showing the amount of products which will be added
after this particular value selection. Counter is recalculated after each filter applying. If some
values become inapplicable after certain filter appliance, it becomes inactive.
5. COMPARISON
There will be a checkbox to add product to the comparison. Comparison should be possible
only for products from one category.
After adding 1
st
product – next to the checkbox, a hover info box appears “Add one more
product to see the comparison”:
After adding 2
nd
product – button “show the compare” appears near both on 1
st
and 2
nd
products:
  • Page 1 1
  • Page 2 2
  • Page 3 3
  • Page 4 4
  • Page 5 5
  • Page 6 6
  • Page 7 7
  • Page 8 8
  • Page 9 9
  • Page 10 10
  • Page 11 11
  • Page 12 12
  • Page 13 13
  • Page 14 14
  • Page 15 15
  • Page 16 16

bintime_test 00000-895612 Product information

Category
Night creams
Type
Product information

Ask a question and I''ll find the answer in the document

Finding information in a document is now easier with AI