Blog → SemanticForce Platform: Creation of Advanced Queries

This is a short review of extra capabilities of search queries on Semantic Force platform, which will show you some other ways of tailoring and detailing sought-after selections. If you like, you can set such search in our system yourself (we give specific examples below) or ask us to make extra settings required.

1. Search in headlines.
The search in headlines allows searching objects only in headlines, rather than in the entire text.
How to use. Using this query you can quickly learn how often your brand is mentioned in headlines of articles. Moreover, you can see, how widely a certain article has spread in the online space, if you know its exact title; or how well a piece of news has played its role, with one or more key words presumed in the headline.
How to set. Create a topic, set main objects of the query (e.g. HTC). Then press “Show Advanced Search” and create a query, e.g. title:"One" Thereby you will find all mentions, where headlines contain a name of a new product line of HTC One smartphones. They will be headlines of articles for online mass media and blogs, and message content etc. for microblogs.

2. Search with distance. It is sometimes not enough to indicate the context of search queries just in words; at times it is also required to specify proximity of explicative words. For such cases one can use a search with indicated distance from one word to the other (the distance is measured by the number of words between them).
How to use. This search query can help in numerous situations. For instance, it helps search mentions of certain types of goods or their features – in those cases context words should be located in the text not far from your brand, and you will see the most relevant feedbacks in search results. The search with distance is useful for such queries as “wanna buy a phone”: there can be usually a few words between “buy” and “phone”: I’ve long wanted to buy a new phone, I planned to buy my wife in the Internet a phone etc. To trace all those queries, it’s worth using the above option.
How to set. Create a topic, set main objects of the query (e.g. Messi in the context of “god”). Then press “Show Advanced Search” and create a query: “Messi god”~4. You will see all mentions, where Messi is compared to god, and are likely to avoid tropes like “football god was in favour of the team” and other mentions of god, which are not related to the person of Lionel Messi. The higher is the number after tilde, the farther are the words from each other.

3. Advanced queries.
The line of advanced search in the system is also suitable to set advanced search queries with standard logical operators.
How to use. You can have a pre-defined search line to monitor an object, so there is no need to divide it into anchor, context and stop words. You just use a pre-defined query on SemanticForce platform.
How to set. Create a topic and add one or more anchor words. Then open a line of advanced search and insert a pre-defined search line, e.g. the below sample query will show all messages, where a battery in Sony Cyber-shot camera is discussed: (Sony && CyberShot && battery) || (Sony && KyberShotШот && battery) || (Sony&& Cyber Shot && battery) || (Sony && CyberShot && battery) || (Sony && KyberShot && battery) || (Sony && Cyber Shot && battery)

4. Advanced endings.
There’s no need to enter a variety of word-form of the object (declension in different cases etc.) manually – it’s enough to create a query with an advanced ending or advanced writing of the middle of a word.
How to use. It makes it easy to insert variations of endings for any words, which is necessary for most queries.
How to set. Create a topic, insert a word for monitoring in nominative case (e.g. model), then write it with an asterisk at the end: model*. Then the platform will give all possible mentions with endings (irrespective of the number of symbols instead of the asterisk): models, modeling etc. If only one symbol after the asterisk is required, use a question-mark instead of the asterisk: model? This query can be used not only for endings of a word, but also in the middle of it: IK?A. The query can be used for all types of words: anchor, context, stop words.

5. Fuzzy search.
Full and exact spelling is usually considered in a query; however a human factor and probable spelling mistakes should be taken into account. This command allows gathering mentions of words similar to the sought-after one. A similarity degree is specified in the query – words can be different both in one letter, and in all letters.
How to use. Mistakes and misprints are usually hard to predict. This query allows considering the most unpredictable nuances in spelling. Moreover, there are rather complicated brand names, which people often misprint. For instance, you can find a lot of mistakes with "Leroy Merlin" brand - Leroi Merlin, Lerua Merlen... How many and what other mistakes can people make? You shouldn’t guess, you can just set a fuzzy search with one or more letters probably misprinted.
How to set. Create a topic, set main objects of the query (e.g., Leroy Merlin). Then press “Show Advanced Search” and set a query, e.g. Leroy~0.6 OR Merlin~0.8 That will mean that the system will search mentions with a probable distortion by 2 letters for Leroy and by 1 letter for Merlin. The index after the tilde can be from 0 to 1; the higher is he number, the more similar will be a word to the original one.

Please be noted that the setting of topics is included in any SemanticForce package of services, and we are always willing to help you in selecting words, context or forming advanced queries. Wish you productive work!

Read more:


Blog comments powered by Disqus