Although this issue has a corresponding issue as a solution(see this issue), but it is definitely not the only or the best solution. Although the midfix of a sentence conveys the most about the relationship between the prefix and postfix, in many unstructured texts, that doesn't, seem to be the case. A majority of texts extracted from social media do not have a fixed schema or structure. This can arise due to several reasons like the inclusion of sarcasm, shorthand typing, or even the lack of grammar knowledge of the user. On the other hand, a significant amount of text scraped from social media handles are of such peculiar forms. Hence, the midfix of a sentence, in every case might not always provide a proper relationship between the prefix and the postfix. The challenge is to find a complete solution, a parameter(or set of parameters) that will most definitely, help deduce the relationships between the prefix(drug) and postfix(ADR) in all cases to enhance the rules-based Labelling Functions(LFs).
Although this issue has a corresponding issue as a solution(see this issue), but it is definitely not the only or the best solution. Although the midfix of a sentence conveys the most about the relationship between the prefix and postfix, in many unstructured texts, that doesn't, seem to be the case. A majority of texts extracted from social media do not have a fixed schema or structure. This can arise due to several reasons like the inclusion of sarcasm, shorthand typing, or even the lack of grammar knowledge of the user. On the other hand, a significant amount of text scraped from social media handles are of such peculiar forms. Hence, the midfix of a sentence, in every case might not always provide a proper relationship between the prefix and the postfix. The challenge is to find a complete solution, a parameter(or set of parameters) that will most definitely, help deduce the relationships between the prefix(drug) and postfix(ADR) in all cases to enhance the rules-based Labelling Functions(LFs).