A suggestion is considered most relevant when it is in the area where the user expects to see suggestions. There are two types of Location Intent which are Explicit and Implicit based on a presence of location intent in the query. Explicit
The query indicates a specific location or area where suggestions are expected.
Query String | Possible Suggestion | Explanation |
โรงแรม เชยี งให | Hotels in Chiang Mai | User and viewport locations should be ignored. |
วตั สนั สมทุ รปรา | The queried chain business within the requested province | User and viewport locations should be ignored. |
รา้ นอาหารเกาะ เกร็ด | Restaurants in Koh Kret | User and viewport locations should be ignored. |
รา้ นอาหารใกลฉ้ นั | Restaurants near the user’s location | Take the user’s location as the location intent. Viewport location should be disregarded, even if fresh. |
If the user did not specify the location intent in the query, use the user location, viewport location, and viewport age to determine the area of expected suggestions.
Fresh Viewport with the User inside
We take the user’s location as the location intent when the user is inside the fresh viewport.
• Suggestions inside the fresh viewport cannot be rated Bad for distance alone.
We consider only the user location as the location intent.
Query: kerry บคุ คโล
User and viewport location near Bang Na, Bangkok
The user requested to see Kerry shops in or near แขวงบคุ คโล (Bukkhalo). We should ignore the viewport and user location. Research shows that there are several Kerry shops in Bukkhalo. Please see figure 1 for the Kerry Express shop search results in this area. Any returned Kerry shop in Bukkhalo would receive the Relevance rating of Excellent. Results in other sub-districts would be demoted depending on the distance from the requested location.
Please also note that we treat temporarily closed businesses as open.
Query: อเม
User Location: 13.6803, 100.4272
Viewport Age: Fresh
Viewport Centre: 13.678902, 100.436246
The user was likely looking for Café Amazon. Since the user is inside the fresh viewport, we consider the user location as the location intent regarding the guideline 2.3.2 Implicit Location section. Research shows that there are multiple Café Amazon stores near the user location that are not returned for rating. We should consider all possible suggestions in the real world. Given that all returned results in figure 2 are Café Amazon. The green result would receive the Relevance rating of Excellent because it is among the closest results to the user. The red and purple results have similar distance from the user but there are several Café Amazon stores closer so the rating of Good or Acceptable are acceptable. The blue result is somewhat significant further away so the rating can be either Bad or Acceptable.
Matching to the query string is the first step in the rating process and is the minimum requirement for a suggestion to be considered relevant. Once a match is confirmed, the relevance rating is rated in terms of prominence and distance. If there is no match between the query and the suggestion, rate the suggestion Relevance as Bad.
General Match
The query string can directly match the suggestion, or it can match the most significant part of the suggestion.
User itent
Some user queries in Autocomplete can be relatively complete strings with a clear intent while others can have multiple interpretations of what the intent could be.
Many schools in Thailand are located nearby temples and share the same name with the temples. Temple school result returned from the queried temple can still be the secondary intent.
Query String | Suggestion | Intent | Highes t Initial Rating | Explanati on |
วดั เขมา ภริ ตาราม | วดั เขมาภริ ตาราม ราชวรวหิ าร | Primary | nEtxcelle | This temple is much likely the user’s primary intent. |
โรงเรยี นวัดเขมา ภริ ตาราม | Second ary | Good (User intent) | The school has the same name as the temple. It matches the user’s secondar y intent. |
The user query is an abbreviation or alternate name, and the suggestion provides the full and currently accurate name.
Query String | Possible Suggestion | Explanation |
scb | ธนาคารไทยพาณชิ ย์ | Match: SCB stands for Siam Commercial Bank. |
dmk | ทา่ อากาศยานดอนเมอื ง | Match: The international airport code for Donmueang International Airport is DMK. It is possible that some users would type IATA code to see a specific |
airport. | ||
sk | โรงเรยี นสวนกหุ ลาบวทิ ยาลยั ุ | Match: SK is an abbreviation of Suankularb Wittayalai School. |
รพสต | โรงพยาบาลสงเสรมสุขภาพ ตาํ บลสาํ โรงเหนือ | Match: รพ.สต. stands for โรง พยาบาลสง่ เสรมิ สขุ ภาพ ํ ตาบล. It is possible that the user simply type “รพสต” instead of “รพ.สต.”. |
o’ โรงพยาบาลสตว | No Match: An abbreviation of โรง พยาบาลสตว is รพส. . |
The user query may be misspelled or contains parts that may have been entered incorrectly, and the suggestion corrects that issue. This type of match should only be used for obvious issues, when no suggestion for the actual query exists at the user location or when the user intent clearly is satisfied by the autocorrected suggestion.
Query String | Possible Suggestion | Explanation |
bys | BTS | Match: Query “bys” is very likely a typo of “BTS” (“y” and “t” are nearby keys). |
โลตส | โลตสั | Match: The query suggestion corrects the user’s obvious mistake. |
เมอื งไทยลสิ ซิ | เมืองไทย แคปปิ ตอล | Match: The suggestion corrects the former name of the business to the recent name. |
The query string could be completed as a category. The suggestion satisfies that potential category.
Query String | Possible Suggestion | Explanation |
้้ รา้ นเค้ | รา้ นเคก้ | Match: The suggestion returns the most likely intent of the user. |
MRT | สถานสี ทุ ธสิ าร | Match: A suggestion returns a specific station that is part of the MRT transit system. |
พซิซ | สกซู ี | Match: The user is looking for pizza. A suggestion for a pizza restaurant is considered a match. |
หางสรรพสนคา | วตสน เซน ทรัลแจง้ วัฒนะ | No Match: A suggestion for a store does not satisfy a category search for “mall”, even if that store is located within a mall. |
When a query address and an address suggestion are not exactly the same, the kind of connection they have depends on their relationship:
Query String | Possible Suggestion | Explanation |
120 ถนน แจง้ วัฒนะ แขวง ทงุ่ สอ | 120 ถนน แจง้ วฒั นะ แขวง ทงุ่ สองหอ้ ง เขต หลกส กรงุ เทพมหานคร | Match: This address suggestion matches the query string exactly. |
ถนนแจง้ วฒั นะ | Match: The suggestion provides only the street name. The relevance is therefore demoted to an initial rating of Acceptable as it is an unlikely secondary intent. | |
แขวงทงุ่ สองหอ้ ง | No Match: The suggestion ignores all street address parts from the query and returns only sub-district. This does not satisfy the user intent and is rated Bad. |
The classification is considered correct if it accurately reflects one of the main businesses, services, or functions of the entity. It can be broad but not misleading.
Suggestion Name | Classification | Name Rating |
สถานบี รกิ าร นํำามันปตท. | ปัม | Correct |
เดอะมอลล์ ราม คําแหง | หางสรรพสนคา | Correct |
นารายณ์พซิ เซ อเรยี | รา้ นอาหาร | Correct |
ฮอทพอท | Correct |
The final name accuracy is always rated Incorrect when the classification is incorrect. Consider the classification incorrect when it is:
Suggestion Name | Classification | Name Rating |
|---|---|---|
เอมเคสกุ | อาหารตะวันตก | Incorrect |
อยธุ ยาวทยาลย | โรงเรยี น | Correct |
โรงเรยี นมัธยม | Correct | |
โรงเรียนประถม | Incorrect |
มหาวทิ ยาลยั | Incorrect |
According to the research อยธุ ยาวทยาลย is grade 7-12 school, a broad classification is still considered Correct whereas misleading classifications such as โรงเรยี นประถม or มหาวทิ ยาลยั for the result อยธุ ยาวทิ ยาลยั are Incorrect.
Business/POI is Closed or Does Not ExistIf the research shows that the business is closed, select the Business/POI is closed/does not exist checkbox.
suggestion Bad because the suggestion would yield no valid suggestions and provide no use to the user.
Please see examples of various evidence sources that can be used to identify
the closed business below.
Street imagery is considered strong evidence and might be the easiest way to confirm that the business is closed when it is available at the result’s location.
We can see an example below that the signboard of Siam Commercial Bank was removed and property for sell sign was placed. Other evidence from street imagery can be the building no longer exist, a signboard was changed to a new business, out of business sign and etc.
Result: ธนาคารไทยพาณชิ ย์ สาขาแจง้ วัฒนะ ซอย 13
Rating: 0 Business/POI is Closed or Does Not Exist
Figure 4 Street imagery from 2012
Post from the Official Social Media
An official announcement is strong evidence to confirm that the business is closed. This type of evidence is likely to be found by searching for the business name along with the branch name on search engines or searching for a branch name on the business’s official social media page.
Result: ชาบู สตรที สาขาอมตะนคร
Rating: ☑ Business/POI is Closed or Does Not Exist
Figure 6 Shabu Street official media post
Businesses and POIs can close for construction, remodeling, vacations, and other reasons. Treat all temporarily closed businesses as open when the closure is announced on the business/POI webpage or managed social media page. There is no limit to how long the closure can be.
Result: Dalha.cafe
Rating: Since the business is temporarily closed, treat the business as open and rate as usual.
Figure 7 Dalha.cafe official media post
Note: This example is for illustrative purposes only, as the business situation is subject to change and this information may become obsolete soon
Store Locator on the Official Website
A store locator page on an official website can be used to confirm if the business still open or closed. An example below shows that there is no result returns when searching for a Karmart store in the southern region.
Result: คารม์ ารท์ โรบนิ สนั โอเชยี่ น นครศรธี รรมราช
Rating: ☑ Business/POI is Closed or Does Not Exist
Figure 8 Karmart store locator