Autocomplete and Search tasks share many characteristics, but it is important to know the differences between the two task types when rating them. In this learning material, we will go through the differences and some typical rating scenarios in the Finnish market.

Here is a list of the main differences between Autocomplete and Search tasks:

1.Autocomplete queries may be complete or partially completed. The Autocomplete feature is intended to save users time by showing them relevant content as they type. When defining user intent, we need to keep in mind that the user might not have finished the query. For example, likely user intents for a query string ”ravi” are ravintola and ravirata etc. Search task queries are always considered complete.
2.In Autocomplete tasks we rate suggestions while in the Search tasks we rate results. Suggestions and results may be demoted for prominence and distance or for unlikely/secondary user intent.
3.The Autocomplete suggestions returned are divided into four types that include name/title and address details:
Address suggestions
Business/POI suggestions
Query suggestions
Category suggestions

Please review the Autocomplete Evaluation Guidelines section 2.2 Suggestion Types for a complete explanation.

The Search results, on the other hand, are divided into three types:

Address results
Business/POI results
Features without an expected address

See the Maps Search Evaluation Guidelines section 2.2 Result Types for a complete explanation.

4.The relevance rating options for Autocomplete suggestions are Excellent, Good, Acceptable and Bad. The relevance rating options for Search results are Navigational, Excellent, Good, Acceptable and Bad.
5.There is no pin rating section in the Autocomplete task, while there is one in the Search task.
6.The expected address components for Autocomplete suggestions (full address and POI results) are street name, street number, unit number (if applicable) and locality. The expected address components for Search results (full address and POI results) are street name, street number, unit number (if applicable), postal code and locality. The expected address components depend on the type of suggestion, but the postal code is only expected in Autocomplete task when the suggestion is simply a postal code. See the Country Specific Guidelines for the complete information.

Please study the Autocomplete Evaluation Guidelines thoroughly to understand all aspects of the Autocomplete task https://geoworkerz.telusinternational.ai/mapping/StartWorking/Docume nts/AC%20Guidelines.pdf

 

Common Rating Scenarios Finland

Example 1. Query suggestion – name is missing a required sign/symbol

Query: smarke

User location: 62.61513622308853, 24.639234195741594

Fresh viewport: 63.42364257735526, 25.012073799163293

Query suggestion: smarket

Correct Relevance rating: Excellent.

Correct Name rating: Partially Correct.

Query suggestions are rated for Relevance and Name Accuracy. This query suggestion is a very likely match to the query but the business name is missing the character “-“. According to the official business web site, correct name would be s-market or S-market. https://tinyurl.com/2bp2qszc The accuracy of query and category suggestions is rated using the guidelines for business/POI name accuracy. One difference is that if a query suggestion appears in all lower case, this is not a reason for demotion. See Guidelines sections 5.2. Suggestion Name and 5.7. Query and Category Suggestions

Example 2. Suggestion in Unexpected Language/Script

Query: Gr

User location: 61.505697686875855, 23.69699248025236

Fresh viewport: 61.49828329163012, 23.732353158524248

Category suggestion: Groceries

Correct rating: Result name/title is in unexpected language or script – TRUE.

In this case the query string is only two letters but a very likely match does exist: Grilli or grillikioski. The category suggestion ”Groceries” is in English and it is not an expected language in the Finnish locale.

A suggestion’s name or title is in an expected language or script when it is:

In the language/script of the test locale
In the language/script of the query
In a language/script of the suggestion’s region
Any combination of any of the languages or scripts described above
An official company, chain, or brand name commonly used in the market, even if this is not any of the languages or scripts described above

Example 3. Relevance: Good – Distance

Query: ab

User location: 62.737315427576654, 28.374890656140686

Fresh viewport: 62.71766780854674, 29.011196522670737

Suggestion 1:

ABC

Name: ABC

Address: Polvijärventie 1, Polvijärvi

Red dot = user location. Red square = Fresh viewport. Blue dot = Suggestion 1.

Correct Relevance rating: Good – Distance. User location is outside fresh viewport, so fresh viewport is the location intent. The most likely user intent is the gas station ABC so we need to research where the closest branches to the viewport are located. On the official web page https://www.abcasemat.fi/asemat we can find a map tool that shows all the branches.

.

We can see that there is one branch located inside the viewport and our result is the second closest real world result. Therefore, it is demoted once for distance. See Guidelines sections

8.3. Chain Business Query and 8.4. Category Query.

Example 4. Relevance: Bad – User Intent

Query: Kajaani

User location: 64.20306371855598, 27.339753854495417

Stale viewport: 63.94284750015068, 26.759347529416157

Suggestion 1:

Prisma Kajaani

Name: Prisma Kajaani

Category: Ruokakauppa

Address: Veturitie 1, Kajaani

Correct Relevance rating: Bad – User Intent. The location intent is the user location as the viewport is stale. There are many matching real world results for the query ”Kajaani”. We can see all existing address matches on https://asiointi.maanmittauslaitos.fi/karttapaikka/ The most likely user intent is the city of Kajaani located very close to the user. There are some secondary matches like ”Kajaanintie” nearby as well.Our result is Prisma Kajaani, a prominent supermarket in the locality. https://www.prisma.fi/myymalat/590772240. However, the query matches only with the location modifier part of the suggestion. The use of “Kajaani” in the business name is extra information and not a part of the official branding for this business chain. This is not considered a match according to the Guidelines section 4.1.1. General Match.

Example 5. Address: Incorrect – Street Number

Query: Scandic Vaas

User location: 63.38076365784365, 26.052384321026295

Stale viewport: 64.24901787924853, 25.275191613640356

Suggestion 1:

Scandic Vaasa

Name: Scandic Vaasa

Category: Hotelli

Address: Rosteninkatu 3, Vaasa

Correct Relevance rating: Excellent

Correct Name rating: Correct

Correct Address rating: Incorrect – Street Number.

The suggestion is the most likely user intent but there is an error in the address details. We can confirm on the official web page that the street number should be 6. https://www.scandichotels.fi/hotellit/suomi/vaasa/scandic-vaasa

 

 

 

Leave a Reply

Your email address will not be published. Required fields are marked *

Related Post