The intention of it is to minimize the dimensions of the brand new enjoy series

For this reason, ReCDroid+ resets the leaf nodes to ready in order to remain the fresh search (Line 19–20)

A great GUI component is recognized as being strongly related this new insect declaration and purchased on the kept of your tree top when next standards is found: (1) it fits brand new bug report and wasn’t looked in earlier in the day levels; (2) abreast of appointment the initial position, it appears to be before on the insect statement according to their related phrase ID; (3) it is a clickable component and won’t meet the first status, but its relevant editable component suits new bug declaration (due to the fact simply from the working out new clickable parts is the fresh new mining bring new application to another monitor); (4) abreast of fulfilling the more than standards, it’s obviously more threatening. The newest implementation considers Ok and you can Over as the obviously more threatening areas (Selecting cuatro), as previous parts is much more planning render the new software to a new monitor. The latest standards (1) and (2) consider the purchase out of S2R for the mining, in order for ReCDroid+ can prevent copy and wrong matching.

New regimen FindSequence (Range 19) establishes and that GUI component to speak about near to pick a meeting series to do next version. If any section in the present forest top are relevant to the brand new bug report, they picks the fresh new leftmost leaf and you can appends they to \( \mathcal \). If none ones section was related, ReCDroid+ traverses the fresh new forest renders off left in order to right until seeking an excellent leaf node that is connected to the new insect declaration. Rather than including backtracking actions to help you \( \mathcal \), ReCDroid+ discovers this new suffix road throughout the leaf to resources are performed next iteration. Whether your formula finds you to definitely none of your leaf nodes was connected to the brand new insect declaration, it means that individuals must deepen the fresh new mining to find out more matching GUI section.

DOET does not bring the become step since it is not good GUI component. As well, of the you’ll destroyed guidance regarding insect report, it is hard to choose the precise location of the change step. Therefore, we have to choose the best places when you look at the a meeting succession so you’re able to type the fresh turn action (Line 4). We fool around with a threshold Roentgen so you’re able to indicate the utmost quantity of steps towards the last knowledge from which rotate are resolved. Trying to find 2 implies that a crash will occurs step one–2 tips pursuing the become. Thus, automatically hot slovenian american women, Roentgen = 2.

step 3.step 3.2 Vibrant Matching.

To determine whether an effective GUI part fits a pest statement (Range eleven), ReCDroid+ utilizes Word2vec , a phrase embedding strategy, to check if the stuff of the GUI role is semantically equivalent which have any of the extracted experience representations or even the terms and conditions out of sentences where grammar designs can’t be used. Word2vec spends a sensory system design understand keyword embedding from an enormous corpus from text. Word2vec means each word of the a mathematical vector. Cosine resemblance rating about range of [0, 1] anywhere between vectors from a few terms and conditions ways the fresh semantic resemblance anywhere between conditions (1 ways an exact meets). The fresh Word2vec model is actually instructed regarding a public dataset text8 which has had sixteen mil words which is offered also the source password out-of Word2vec . This new model uses a rating regarding listing of [0, 1] to suggest the degree of semantic similarity between terminology (step 1 suggests an accurate suits). ReCDroid+ spends a relatively large get, 0.8, given that threshold. I seen one to playing with a low tolerance may mislead the fresh search for the a wrong GUI part. Such as for example, new resemblance get from “start” and you can “stop” try 0.51 however the a couple of words aren’t synonymous.

0 respostas

Deixe uma resposta

Quer juntar-se a discussão?
Sinta-se à vontade para contribuir!

Deixe uma resposta

O seu endereço de e-mail não será publicado. Campos obrigatórios são marcados com *