Malaysia – Autocomplete common mistakes

Important Guideline Sections

This learning guide aims to highlight several common mistake trends among analysts in Malaysia. Please review the following guideline sections as it will help you rate more accurately.

Autocomplete Guidelines

NOTE: Guideline sections in this guide refer to the May 2024 Autocomplete Guidelines.
Refer to the latest guidelines if there’s any conflict between this guide and newer guidelines.

1. Research & Context

.

Research is important to discover all possible interpretations for the query:

At first glance, it appears as if the suggestion has no relevance to the query. However, research shows that this is a Poslaju pick-up and drop-off point, and is therefore relevant. (https://is.gd/KcNb60) (https://is.gd/fmfiQA)

Other mapping resources also provide context. Nearby function > Search for poslaju. There’s at least 5 closer possible results to the VP and the suggestion is further away from the VP. (https://is.gd/9QzO8l)

Considering the above explanation, Relevance is Bad for distance.

See GL Section 2.1.1 Research Expectation and 4.3.2 Distance.

1.Research & Context (continued)

Context from other mapping resources (https://is.gd/9QzO8l)

.

.

Query is for a location (Melaka). VP is located at an odd location (0,0). There are no possibilities anywhere near the VP.

(Melaka) can be treated as explicit location intent. VP and user locations will therefore be ignored.

The suggestion (Dataran Pahlawan Melaka Megamall) only matches the query string (Melaka) on the location modifier part of the POI name, which is nonessential. This should not be considered a match.

Suggestions such as (Melaka) and (Melaka Sentral) on the other hand would be considered as matches with the query. They meet primary/secondary intent depending on how prominence is weighted.

Considering the above explanation, Relevance is Bad.

See GL Section 2.3.1 Explicit Location and 4.2 User Intent (holywoo) example.

.

The query is for (lorong). Since the user doesn’t specify the exact ‘lorong’ (lane), any ‘lorong’ could be considered a possibility, subject to distance from the user.

Research shows that there are numerous ‘lorong‘ closer to the user (https://is.gd/kDD0T1) (see next slide). Therefore, the suggestion will be demoted by 1-2 for distance, and Relevance is Good/Acceptable.

See GL Sections 2.3.2 Implicit Location and 4.3.2 Distance.

.

.

Context from other mapping resources (https://is.gd/kDD0T1)

4.Few Possible Suggestions (continued)

5. Distance & Prominence

Query is (pandan). Research suggests that the user may be looking for

 nearby businesses (Pandan Republic), transit POIs (Pandan Jaya LRT), or sub-locations such as (Pandan Indah, Pandan Jaya). (https://is.gd/D7ruUQ) (see next slide)

Considering (Pandan Jaya LRT)’s prominence and distance from the user, it could be one of the user’s secondary intents.

Given the above explanation, Relevance is Good.

See GL Section 4.3 Distance and Prominence including subsections.

Note: While there are several closer POIs with (pandan) in their name e.g. (McDonalds Desa Pandan, Nasi Ayam Bonda Desa Pandan), they aren’t necessarily a match with the query. See GL Section 4.2 User Intent (holywoo) example.

.

.

5. Distance & Prominence (continued)

Query is (pandan). Research shows that the user may be looking for

nearby businesses (Pandan Republic), transit POIs (Pandan Jaya LRT), or sub-locations such as (Pandan Indah, Pandan Jaya).

(https://is.qd/D7ruUQ) (see previous slide)

Note that the suggestion is a category suggestion. Category suggestions do not appear on the map, but are still rated according to specific criteria per guidelines.

(Pandan Chicken) is a type of food dish, however it is not very prominent when compared to the other possibilities mentioned above. Furthermore, it is more likely one would search for (Thai Cuisine) rather

than (Pandan Chicken) specifically.

Considering the above explanation, Relevance is Acceptable/Bad.

See GL Section 2.2.2 Query and Category Suggestions and 4.3.4 Distance vs. Prominence.

7. Query Suggestions

Query can be interpreted as a category query for clinics. Since the VP is stale, distance will be determined from user location.

Note that the suggestion is a query suggestion. Query suggestions do not appear on the map, but are still rated per guidelines criteria.

Research shows there’s a (Klinik Alam Medic) ~2.4 km from the user.

Since this is a chain business, it is accepted as a valid query suggestion.

There are closer possibilities (Poliklinik Mas, Klinik Mediviron Taman Mas) ~500m from the user, (Klinik Fatiha Medic) ~1.6km from the user. (https://is.gd/x6k4of) (see next slide)

Considering the above explanation, demote for distance. Suggestion is within the 3^rd closest range from the user. Relevance is

Acceptable/Good. See GL Sections 2.2.2 Query and Category Suggestions and 8.4 Category Query.

Context from other mapping resources (https://is.gd/x6k4of)

Recap