████    重点词汇
████    难点词汇
████    生僻词
████    词组 & 惯用语

[学习本文需要基础词汇量:9,000 ]

[Music]

[Music]

yeah it's okay presentation so the

current title is and approaches to

testing future payment scheme software

development things my name is Daniel and

my supervisor is Civic of cheese so

talking about the context the

identification is that the motivation of

the thesis is that the fictional

comments testing is very potent that

area of the ignore it and they should be

studied further so the aim of the thesis

is that how companies deal with a new

feature commence testing in software

development things also did a method of

the thesis will will have a literature

review and also a survey so the goal of

the typical study you'll need to

understand approaches and challenges of

testing movement requirements the point

of view of software development team

member in the context of organization

developing the development software so

as chair achieve my goal

I need your answer then I have some

questions that with the answer of those

questions I can enrich people so the

questions are what types of content

requirements are specified in a software

project product as the tested and valued

also how much control requirements

testing is done by the type of

in a software project our products what

are the challenges in which recommends

testing by testing like type of

infection requirement in a software

project or product and at least what

characteristics is not for our common

testing infested by in the software

project predicting okay may I have a

question sorry for interrupting years

about a because you are using frequently

the word type of if ours is so it would

be nice if you could define base earlier

right before formulating the goal you

saying what types of anaphors are the

what is done by type of NFR and so on by

type by type so yeah sorry for

interrupting but he said anything no

time would you please explain what is

type of an ephah or for you yes I'm

explaining during the next slide that

internationally it should be before

episode so as you can see there the type

I mean start over the first question

what types of non-functional parents are

specified best bedded in a software

product team so the metrics they are you

I use she's answered this question will

be the following types of control panels

that they are defined by their eyes own

25,000 ten so they are a types of

categories such as accessibility

performance and so on and also I will

have the answers from a survey they will

use

Likert scale five points yep also

intuitive analysis that will be based

under the number of responses and also

the person page comparing to other types

of answers so it but to make it clear

okay why you are talking about eight

types in isolate a twenty five thousand

we have so-called characteristics and

some characteristics yes so this is the

characteristics I probably choose

another is that you should be

characteristics characteristics okay and

you know my impression is that not a

each big characteristic is suitable for

testing some of them are perhaps no

suitable for even an analysis recent

programs you know programs winters

ability ability so it's difficult to do

it by testing then winter can tell you

that you are using a specific very

specific and clear definition like for

instance NC as you know in some

industries like motive not all industry

they do not allow to use int you have to

specify in sixteen there is such as

undock visitor sixteen and some winters

are checking those characteristics so

and it would be nice to see a if you are

taking all those even I

to go down to sub characteristics I love

them are you going to get

the response from industrial industrial

people question we'll be looking right

so your question particular question

will be what types of a farce

are you specifying in your company are

you testing yes I have enough that the

final survey but but I have in this

prototype I am asking if the type of

enough are is being specified on the

project or product you are working and

if it's specified and and if it's tested

how the approach for testing it it could

be for example for tourism use you sleep

for example this is a valid answer okay

so I like to see very much your

questionnaire okay I know okay so I

understood your message our chart that

you have results but you even don't have

design all those questionnaires is not

ready okay the second question is hope

and a for testing is done by type of and

a for is perfect to answer this question

now you use few metrics so again the

type of functional panels will ensure

planet I pray that if analysis it will

be for example the categorization of

type of approach for example if you are

using a spectral running or if you are

using tools or maybe even specific rules

of course based on SO shopping

and then it's okay you can open is the

same yes again even in testing by type

of the control panel use of the project

or product again the type of percentage

and also discovery of new changes also

based on text answers you know that

question it would be very good when put

in the context of results of your

systematic initial read you a so it

could be done in the following way a in

going through a literature you have

identified sound charlie

and then you could verify those

challenges by asking a professionals

it's a really concerned that this is a

challenge I have a question in the

survey that it's it's basically the

water talking about I have a question

with every option is a challenge that I

I found only in a church organ and then

a person can choose and also add more

okay okay okay sounds good so the first

metric that the fourth fourth question

gets the last one what characteristics

in functional testing are infested by in

a software project our product team so

it's an effort testing I don't get it

so yeah I think then this question is

certainly written so it's so it's what

Curtis are increased by in soil testing

so what what is its influences testing

influence influence characteristics

influence testing maybe which

characteristics less suitable for

testing so types of components analysis

and also the court a sextet will be

investigated that can influence or not

fantastic for example we have here

country will approach as methodology

that assisted the main experience we

could try to identify for example some

relation between experience for example

if persons that work with IT from for

wartime they tend to pass the morning

forest or the size of organization

bigger organizations they should that

estimate types of payments and see the

value in them for independent variable

in other words in other words the

question is that's an affair testing

depends the country and other factors

yes okay okay yeah and you will be

looking for in something that it is

specific beauty to a country or region

yes exactly to the domain or experience

or size of organization some people are

doing this okay so it's a question about

independent variables influencing

approach to NFR tasting okay okay thank

you so yeah for procedure and I'm still

although these three questions

designing the the the survey questions

and after we choose a to selection also

strategy to reach population benefit

analysis then try for a pilot study get

feedback about the passion area then

improve the questionnaire then finally I

will in spread it you spread the idea

the final version also collect the

answers and analyze data then by the end

of the analyze the data I will create my

report based on conclusions okay so yeah

the depression area that I'm creating

I have to be very careful about the

balance between the refreshes also about

the priests questions because they are

easy to feel but they don't really they

don't have more specific details and

it's the opposite of open questions that

they are trying to fill but they they

have more specific procedures and also

list these questions with facilities job

options to make list that the person who

is writing the quiz feeling that the

survey cannot test is to use instruments

I said previously that I I didn't choose

but actually we will choose the service

because they can they they have a very

good way to create service we had number

of questions and also the results will

be exporter exported directly in CSV and

will be easy to migrate for example show

the tools that I use to analyze data so

I have changed from Becca because I have

disgusted with the neutering we thought

about those two options are or the

library for Python and us and we didn't

choose yet which one I will have to

research for them

so chocolatey friends I have the

constructibility so the initial set of

challenges can be understood differently

by different people and yeah it strategy

it will actually be except this because

there is no there is no way to to

control this but as a potential value

mitigation strategy will be to double

qualitative analysis first by me and

then my by my supervisor also will

fellow already before deadlines sure

this bias is in canopied the

trustworthiness of respondents there is

pellets would be a by should should feel

that we failed and Indians of the survey

there will be a field to fill the email

maturity results so I think this

strategy will if I receive a lot of Neos

probably the this survey of the specific

person will have a more that trust in

their answers that not be able to

determine who received a know many

people got the invitation this is there

is no really mitigation for these I

think I will just accept of course I

will try to reduce at most enough

schemes of respondents so yeah yes

that the our food enough in English and

also sees it's a review about testing

the control panels we we believe that

persons who are P Phi T have the

capabilities of feeding online form and

as exam about their validity as the

first Japan's representativeness of

respondents and in coverage of different

context by if you can

[Music]

so I may have two prototypes so I show

the second version discussing Oh enjoys

with us great hello Angie can you make

it a bit bigger I can't read this

yes this prototype is each page I'm

talking about the type of function

planet and for each integer data type

I will ask for patience so do test this

type of electronic and it's indicated in

the sections of this one is community

your project or product so this is the

first rule is a specified no.1 yes and

the next question if it is specified

sorry if it is tested so here should

last for all how do you test it which

approach to use also both challenges of

testing

[Music]

[Music]

[Music]

[Music]

[Music]

[Music]

[Music]

[Music]

okay okay okay

then I will I will try to send you some

remarks and feedback okay thank you very

much for your presentation now let's

switch to do n so when and now I will

make you I will make you the co-chair of

this meeting the co-host okay first of

all I have to switch down you make

withdraw and now I can assign this to

so ask your cat can't share my screen

it's very surprising to me that I can

make Daniel Purcell sting the speaking

but not you all right please give it

back to me

after your presentation all right okay

now we are the hosts okay thank you

so Kashyap take some ice cream yes okay

oh yeah so hello everybody and today is

ethical studied for my master's thesis

and my promoter is ninja Moshe and my

the title of my thesis is geometrist

algorithms for scheduling a related

machine the way the network turret

criteria and first let's have a real

about shop my motivation the motivation

units it's difficult but hip virtue so

scary improver in a complex workshop

environment and my aim is to design

effective algorithm to solve the

schedule improver a multiple unrelated

machines with the goal of total Network

minimization under my method at first

I will first I wrote design major

heuristic algorithms and then I will

generate data

said for for my my scheduling problem

multiple unrelated of machines and then

I will contact an experiment to test

differences between event algorithm and

retina at last I will analyze the

results the reels of the experiment and

make by conquering okay left is a

empirical and my goal is to divide age

and compare better heuristic algorithms

we made them total network unrelated

machines and I focus on the stakeholder

could be some ideas of the factory work

workshop who is responsible for solving

for solving scaling problems with

machines and the context factor could be

specific factually workshop and because

my my goal is to measure and to compare

and different material sick I've ridden

so here are three questions for my so my

goal and the first one is a maybe I well

I have changed a bit and I could I could

I change a another's right because I

party I changed something it's not to

the correct flight sorry looks pretty

good

okay okay it's gone yeah nothing I so

sorry sir mm-hmm so the first question

is what what is the accuracy of that

event or that event permit eristic

algorithms actually at first to the

question if a watch is a scale of the

problem of the designed a method can

herself and because professor Ortiz and

me the comments as well I would like to

change yes oh so now I can explain that

to this question and I remember you

asked me and you think it's not it's not

it's not a Staffordshire mayor if if

input data yes if it's a part of input

data but is also the message I want you

mayor because every evidence cans of

different different size of data set so

if the scale is bigger which means the

performance of the algorithm is is

better so I need to because this light

looks okay to me standing I was having a

different slide or slides

in line giggles yeah okay okay thank you

so I'll go up okay that the second

question is which agreement and gets

your best projecting function value and

the last question is how much does

designed algorithms improve the initial

solution about to the initial solution

and that's because and almost every

legend heuristic Armada has a initial

solution to the to the beginning add to

the initial stage so we can compare it

and compare metaheuristic algorithms in

terms of how much they have introduced

to the initial solution okay so here are

also straight metrics for the questions

of both the first metric is number of

machines under tasks and injury we want

a method to solve problems on the larger

scale the better so under this metric is

a message the methodological hangzhou

sample had turned handle 100 jobs either

better than the methodological only

hundred ten jobs the messages carry on

five machines is better than their

method that can only run on one machine

but this is my remark so number of

machine instance is not what you are

measuring you are designing as far as an

I must understand you're deciding a

plant

what is the number of machines that are

available

[Music]

[Music]

[Music]

[Music]

okay so could I replace this matrix into

something like a per Bunch or an

emulation an imitation of the algorithms

yeah if either watch I want you I want

to see sorry so what is the definition

of value of total light work what is the

definition of this value about this this

metric is is it's the optimization and

objective value it is what it is a

metric that we want you

so we must know okay so baby there is

not one is just I mean the value of the

object to the function one do not mean

the value of total network is the object

if I want to which my so maybe what is

this object is we can again define this

in very different way objective is not

one thing right you can have different

thing if ou you objective can be the

number of spittle

it was behind okay so I want to show

another another slide about return to

the previous one is return to the

previous because it's not clear to me

what you mean by increasing the rate to

initial solution again I would expect

that you will define this precisely okay

about the increasing rate in it's

because almost two each so baby I can

and talk to the flights and later slide

to expressive magic power 2/3 to $1

right like this right and this is a

comma free of of corner ledger heuristic

younger of the far left of you can say

almost and

and there are some specific majority of

on the right as you can see almost all

other measures metaheuristic has has one

of my initial solution to the beginning

idolater the initial the initial

solution could be improved by by running

by own image heuristic algorithm so so

so if so I'd like to barrier the

increasing rate to measure the

performance of these algorithms I'm very

sorry but it's not perhaps true that a

metaheuristics is always giving you a

better solution metaheuristic have this

ability that is going outside of the

local search right so it can happen that

the next solution it will be worse but

the next one will be better and so on so

it's not a monotonic a research it can't

be because of the nature of the

complicated this is the nature of meta

heuristics and this is giving chance to

get better results than classical

optimization yeah so do you mean the if

not profits you said to the increase the

rate here I'm just saying that this is

not obvious and you can define and you

should define this so I would introduce

something like I don't know improvement

rate and this improvement rate could be

measured for instance average time that

you have to do

you jerk in order to find next point

which gives you better result it will be

distancing the number of which I would

say jumps right so we are going to

another place worse another still not

better not better then I have MC test

steps were necessary in order to get

better result okay and that would be a

different definition of this so how fast

we are getting a results that it should

be precisely defined that way that you

are showing the slide you see we have

initially some solution and still

defined yes because if not they were

described here so maybe your your RV too

confusing so now and do I understand why

I put two these magic here okay so you

know what to do right yeah okay so these

because we are running out of time okay

so there are three questions and three

metrics in total here and and every

metric increase and corresponds to one

question okay so when I'm not sure and

instead of putting question 1 you should

exactly put the text of the question

number of machines and tasks and

proteins and metrics the metric for it

is the same as you what is the question

[Music]

so so the first match the question is

the either what first get wasn't scales

the data said to the program each other

a great question one is about accuracy

okay accuracy what you had the metrics

to accuracy the currency is a second

metric in a totally to watch it was

question number one sorry yes sorry I

lost to my another slice okay so next

like okay you know what to do so the

next like a procedure a procedure of a

favorite framework of metalist

agreements for Soviet and travel sales

stories sorry when it's a

misunderstanding

I guess that for you procedure it is

computer procedure no procedure here is

a procedure for experiment okay so

you're doing something like writing code

then preparing next step would be

preparing input date and selecting

workbench or whatever then running

something like 100 experiments and

taking a matrix and then computing the

average or something like business all

right this is the procedure that will be

executed by you step by step what are

the steps in your procedure but here

what I see it is procedure for your

optimization program just outline very

general and is saying almost nothing

because we know this general right but

the question is how we are going to

perform your experiment okay so maybe in

the

okay so I so I hope so

[Music]

[Music]

okay

the basis of let's work okay so I'm very

interested in this and thank you very

much everybody for attending this

meeting see you in two weeks time

bye-bye and special bye thank you bye

bye

hi Daniel

I'm here how you here

what and I talk to you

sorry I I don't catch you I just I want

chat with you but if you don't have time

we can end

right here