Gent. It seems to be the trend for many HRM related applications to introduce the feature of ‚related concepts‘ during the process of searching. In other words : while typing in a specific keyword, the application suggests a list of synonyms or related concepts. Related concepts are not synonyms but concepts that have some kind of relation to what you are looking for such as ’sales‘ and ‚international sales‘, ‚recruitment‘ and ’staffing‘,…
Synonyms and related concepts are supposed to return better search results. But…is this how it really works?
Let’s have a look: How do applications define related concepts?
Many related concepts are defined on a statistical basis : if users tend to use concepts in the same context and if many users are doing this, then the application will assume that these are related concepts. This can be a good approach in some cases but it not always is. Let’s take an example : someone looking for .NET development jobs, can type in .NET, Object Oriented,…. If many people do so, then the system will start to assume that .NET and ‚Object Oriented‘ are related. This is fine but when using these as related concepts, will you get better and more accurate results? Certainly not : Java is also related to ‚Object Oriented‘ and while searching for .NET, one doesn’t want to get results like ‚Java‘ …
Another example that is found frequently: when you are looking for sales people, applications often return people with marketing related skills. Sales and marketing might stand in some relation to each other, but they are definitely not to be considered as synonyms!
In other words, we need to be careful when using statistically generated ‚related concepts‘ as these can lead to less satisfactory results – or even the opposite of what is intended! Statistics are just not accurate enough as these do not include the meaning of the words. Specifically: wrong expansions lead to wrong keywords.
The better Approach – Validation of related Concepts based on Meaning
Statistics can help and for sure : recruitment, staffing services, permanent placement, screening services, IT recruitment,…. (related concepts generated in the skills option of LinkedIN) have something in common. But using these will not (always) lead to better results.
The answer on how to define ‚related concepts‘ that really help, is by including the meaning of the words and by using different ways to use the related concepts : in some ways ‚all related‘ can help. Let’s take the example of .NET, Java and ‚Object Oriented‘. While searching for ‚Object Oriented‘, Java and .NET should pop-up – while searching for ‚.NET‘, Java shouldn’t…
Intelligence is required in order to validate and understand the broader meaning of related concepts!
Actonomy’s related Concepts for Skills and Expertise
The above examples illustrate that it is not sufficient to only use statistics in order to define valuable related concepts. Hence, use the ’skills expansion‘ feature of Actonomy xMP, thinking for you in an intelligent way instead of using too general concepts. More intelligence and more rules that really control the meaning of the words: that is the basis for qualitative skills expansion. This is exactly why Actonomy xMP is using rules that define the related concepts depending on the context in which one is searching : while searching for ‚Java‘, you also want to find ‚Object Oriented‘ but not ‚.NET‘ – while searching for ‚Object Oriented‘ you want both ‚Java‘ and ‚.NET‘.
Contact Actonomy to find out how expansion of skills can improve your application. Skills expansion is one of the new (improved) features of xMP 4.0.
B – 9000 Gent