IV FMA 2018 - Work Session IX
This is a modal window.
Das Video konnte nicht geladen werden, da entweder ein Server- oder Netzwerkfehler auftrat oder das Format nicht unterstützt wird.
Formale Metadaten
| Titel |
| |
| Serientitel | ||
| Anzahl der Teile | 13 | |
| Autor | ||
| Mitwirkende | ||
| Lizenz | CC-Namensnennung 3.0 Unported: Sie dürfen das Werk bzw. den Inhalt zu jedem legalen Zweck nutzen, verändern und in unveränderter oder veränderter Form vervielfältigen, verbreiten und öffentlich zugänglich machen, sofern Sie den Namen des Autors/Rechteinhabers in der von ihm festgelegten Weise nennen. | |
| Identifikatoren | 10.5446/48207 (DOI) | |
| Herausgeber | ||
| Erscheinungsjahr | ||
| Sprache |
Inhaltliche Metadaten
| Fachgebiet | ||
| Genre | ||
| Abstract |
|
5
6
7
8
9
10
11
12
00:00
ArbeitszimmerRaumstrukturArchitekturStadtplatzErdbauStädtebauUrbanitätBetonbrückeArchitektComputeranimationVorlesung/Konferenz
01:30
RaumtemperaturVorlesung/KonferenzBesprechung/Interview
02:46
Vorlesung/Konferenz
03:55
PrivatgrundstückVorlesung/Konferenz
05:05
KunststoffrasenVorlesung/Konferenz
06:19
ErdbauMiniaturmodellKunststoffrasenPrivatgrundstückVorlesung/Konferenz
07:25
PräfigurationWohndichteMiniaturmodellVorlesung/Konferenz
08:44
MiniaturmodellÜberdachungPrivatgrundstückVorlesung/Konferenz
10:16
SäuleÜberdachungPrivatgrundstückVorlesung/Konferenz
12:31
SäuleBauteilProfilblechMiniaturmodellRaumstrukturVorlesung/Konferenz
15:10
PrivatgrundstückRaumstrukturVorlesung/KonferenzBesprechung/Interview
16:27
MiniaturmodellRaumstrukturUmbauter RaumPrivatgrundstückMiniaturmodellVorlesung/Konferenz
17:38
ArchitekturMiniaturmodellÜberdachungVorlesung/Konferenz
18:50
ArchitektGeoinformationssystemVorlesung/Konferenz
20:02
HolzrahmenbauPrivatgrundstückVorlesung/Konferenz
21:11
PrivatgrundstückGeoinformationssystemVerkehrsstraßeVorlesung/Konferenz
22:22
SäulePrivatgrundstückCluster <Wirtschaft>MiniaturmodellVerkehrsstraßeSBR-VerfahrenVorlesung/Konferenz
24:11
SchiebfensterBauteilGrundrissBesprechung/InterviewVorlesung/Konferenz
27:03
PrivatgrundstückRaumstrukturErdbauBesprechung/InterviewVorlesung/Konferenz
29:59
PrivatgrundstückBesprechung/InterviewVorlesung/Konferenz
32:44
ArchitekturWasserdurchlässigkeitUrbanitätBauträgerArbeitszimmerKaufhausÖffentlicher RaumWohndichteLouis-quinzeVorlesung/KonferenzBesprechung/Interview
34:17
ProfilblechUrbanitätUmlandVerkehrsstraßeWohndichteCityFußgängerzoneWärmespeicherungArbeitszimmerGeschäftsviertelÖffentlicher RaumVorlesung/Konferenz
35:25
GebäudeGeschoss <Bauwesen>CityArbeitszimmerRolloGrünanlageMobilheimHausWohngebietGeschosswohnungsbauVerkehrsstraßePrivatgrundstückPräfigurationSchwach radioaktiver AbfallUrbanitätVorlesung/Konferenz
36:37
BaublockPrivatgrundstückWasserdurchlässigkeitMauerGebäudeGeschoss <Bauwesen>VerkehrsstraßeRolloGrünanlageHauseigentumFlächenverbrauchPräfigurationRaumstrukturVorlesung/Konferenz
37:49
RaumstrukturÖffentlicher RaumWasserdurchlässigkeitArbeitszimmerHauseigentumPräfigurationMiniaturmodellFlächenverbrauchVorlesung/Konferenz
39:16
RaumstrukturGeoinformationssystemPrivatgrundstückWasserdurchlässigkeitKünstlervereinigungGrundrissLärmschutzwandTürGebäudeWasserwaageAutobahnbauÖffentlicher RaumVorlesung/Konferenz
42:27
EinkaufszentrumCityUmlandWohndichteRaumstrukturWasserwaageHausPrivatgrundstückVerkehrsstraßeRegionÖffentlicher RaumUrbanitätVorlesung/Konferenz
45:34
WohndichteWasserdurchlässigkeitUrbanitätSäulenordnungFlächenverbrauchPrivatgrundstückVerkehrsstraßeVorlesung/Konferenz
46:44
WohndichtePrivatgrundstückÖffentlicher RaumVorlesung/Konferenz
47:56
WasserdurchlässigkeitGeschosswohnungsbauPräfigurationWohngebietPrivatgrundstückVorlesung/Konferenz
49:02
PrivatgrundstückÖffentlicher RaumWohngebietWasserdurchlässigkeitGebäudeVerkehrsstraßeTrennkanalisationRaumstrukturKommunalplanungMauerRolloMiniaturmodellEmpireWohndichteCityPräfigurationGeschosswohnungsbauUrbanitätArbeitszimmerFlächenverbrauchSchwach radioaktiver AbfallGrünanlageErdbauVorlesung/Konferenz
55:25
UrbanitätArchitekturTraggerüstErdbauRaumstrukturArbeitszimmerPrivatgrundstückInfrastrukturWeltstadtGebäudeVorlesung/Konferenz
56:53
GebäudeInfrastrukturArbeitszimmerPräfigurationSchwach radioaktiver AbfallVersorgungsnetzVorlesung/Konferenz
58:04
BauträgerCitySäulenordnungRaumstrukturMiniaturmodellVorlesung/Konferenz
59:32
KlimaanlageCityWasserwaageGrundrissVorlesung/Konferenz
01:00:49
HauseigentumWasserwaageCityUmweltplanungSchlusssteinEmissionKonsistenzVorlesung/Konferenz
01:03:46
GebäudeTraggerüstPrivatgrundstückRaumstrukturEmissionWasserwaageCityErdbauTürBesprechung/InterviewVorlesung/Konferenz
01:06:25
BaufälligkeitTraggerüstPräfigurationRaumstrukturFuturismusVorlesung/KonferenzBesprechung/Interview
01:07:36
StadtplatzPrivatgrundstückRaumstrukturVerkehrsstraßeCityWohndichteProfilblechVorlesung/KonferenzBesprechung/Interview
01:09:57
PrivatgrundstückPräfigurationGebäudeUrbanitätVerkehrsstraßeWasserdurchlässigkeitHausMauerVorlesung/KonferenzBesprechung/Interview
01:12:34
PräfigurationGebäudeWohndichteRolloHausMauerVorlesung/KonferenzBesprechung/Interview
01:13:41
GebäudePräfigurationWasserdurchlässigkeitWohndichteBesprechung/InterviewVorlesung/Konferenz
01:14:54
PrivatgrundstückMauerWasserdurchlässigkeitGeschoss <Bauwesen>HausGeschosswohnungsbauArbeitszimmerPräfigurationBesprechung/InterviewVorlesung/Konferenz
01:16:46
ViaduktErdbauPrivatgrundstückArchitekturBesprechung/InterviewVorlesung/Konferenz
01:19:16
ArchitekturEmissionMiniaturmodellVorlesung/KonferenzBesprechung/Interview
01:20:29
SpätimpressionismusArchitekturBesprechung/Interview
01:21:35
ArchitekturSchützFuturismusBohrinselSozialer WohnungsbauRaumstrukturErdbauCityBesprechung/InterviewVorlesung/Konferenz
01:22:47
ArchitekturBauträgerVorlesung/KonferenzBesprechung/Interview
01:23:56
Besprechung/Interview
01:25:07
ErdbauSchützBesprechung/Interview
01:26:15
GrundrissArchitekturSpätimpressionismusBesprechung/InterviewVorlesung/Konferenz
01:27:37
Organischer AbfallBesprechung/InterviewComputeranimation
Transkript: Englisch(automatisch erzeugt)
00:00
We are beginning our session number nine and last. First presentation will be from Jean Ventour, he is an architect and PhD candidate in Ischter,
00:27
Lisbon, supported by FCT grant. The work of this test is explore teams focused on the dichotomy between emergence and composition in the urban setting.
00:40
The multivariate analysis of urban morphology configuration and use of public open space with relevance to the Portuguese mainland urban square. Exploring the potential of computational tools and data mining and machine learning techniques, it tests their ability to embrace the complexity of those relations and to
01:04
bridge spatial formal analysis and urban design. The litigation of his investigation has been carried out through national, international presentations, publications and workshops. He graduated from the faculty of Porto University in 1998 and exercised since then professional
01:23
activity on architectural and urban design, both in collaboration as individually. Post graduated in 2013 from the course of advanced studies in digital architecture in Ischter and the faculty of Porto.
01:41
He is making presentation on data mining, spatial analysis and algorithmic design. I will talk about a little bit of the subjects that we talk there, mainly about data mining, but in a very practical way, introducing the concepts, some tools, some methods and
02:10
then to illustrate with some exercises. Mainly it was data to be made for every, there was no, and I assume that the participants
02:21
know little or nothing about the data mining, so I gave a review on the subject in some contextualization. This was just the outline of the workshop. When they workshop, they were eight people there, it was nice, very, very relaxed, ambient
02:40
and mainly in the morning we just introduced to the concepts and tools and in the afternoon we play a little bit with them, in this duality, first data mining and then Python scripting and trying to connect the two.
03:04
Now for the introduction, I just say that, I start to introduce it and just saying that it becomes such a big word in the last time, for some time now, since the upscale
03:20
of the internet and the big bad effect and, but even in more recent times, in a few weeks, it becomes, it was brought to the highlights even for not for the better reasons but I was interested to show them what, these are the Cambridge Analytics website,
03:47
before the workshop I went there to see it, I never saw it and it was interesting to see Cambridge commercial and political side by side, like it's the same thing. Of course, about the news, about the relation with Facebook and in the Trump, in the President Trump campaign,
04:08
it says it's a data-driven digital campaign, it's something. And the other, at the same, almost at the same time, it was the first mortal accident with a self-driving car in the United States and the rage against the machine
04:26
and that it worked in the Californians and so on. But just to contextualise, there are a lot of people studying these effects on humankind in a serious way, so for instance his book of Niko Kostrom, of superintelligence,
04:46
in many they are concerned with the point in time where machine intelligence will superpass the human intellect and they call it the singularity and they point it out like for 2045
05:05
that will be suppressed by the machines. Now, after this context, I just introduced data mining as a subfield of computer science that connects with other subfields, also has several names depending on the field and where it's applied.
05:28
Data mining has a more connotation with business and marketing and so on, but they use the same tools as other fields in technology and science, so there is everything around us.
05:43
Now, in a certain point, it uses data mining or machine learning more properly, the field of computer intelligence, artificial intelligence that deals with learning. So, there are several names for data mining and
06:03
when they talk about the special, they're a kind of a special data mining also, but I think a good mathematician would say it's all statistics and the rest is just... Data mining in itself is a standard process or procedure that starts with the
06:24
definition of the problem, you have to understand it, identify the required data, the required data and acquire it, prepare it in pre-processing, this phase it's normally about 85% of the work and then model the data would be applying machine learning algorithms to model it
06:44
and then train and testing, they are connected, so to validate your model and then deploy it or use it or present it, this is a standard procedure and data mining can be
07:03
understood as the practical application of machine learning. Just to conceptualize, I was saying that learning is a field of artificial intelligence, but artificial intelligence has a lot of goals, so besides reasoning, learning would be one of them,
07:23
so this is the part where the mining works with more, much an induction thing from you know, it's just a reduction thing to learn from the bottom, you would say, from examples.
07:41
In learning, in the machine learning, there are two types of learning, I would say supervised learning and supervised learning, these are when you know what you are trying to classify or when you don't have an idea what you are trying to classify or you are working with like images or plain text, something like this, there are other reinforcement learning,
08:04
but these are the two main, when you are talking about learning and machine learning, there are only these two kinds, supervised, you know you are classifying or making a good regression or unsupervised, in this case if you say unsupervised you are doing classification
08:20
or density estimation or something like this, or reducing the dimensionality, then I just introduce the difference between classification and regression and this is just an example for illustrating what is the difference between modeling as a
08:44
regression line or modeling as a decision boundary or something like this, between to classify and to make a regression, if you are that it is continuous or it is not continuous. And then the overall main scheme for machine learning, you have the data normally and you
09:08
split it in two parts, one for testing, one for training, you make a model on the training part, you make prediction on the text part and you make then the evaluation with the labels of
09:23
the data that never entered into the model construction, so it is just a generalization of the process, then the approaches can be a lot of them, there are approaches or algorithms for that several in depending on the case, you could use one or another, normally you use several
09:44
several at the same time and then you classify them and you try to use the best one and you do it, here it was just for illustrating a case of this Latin Semantic analysis, this will be automatically classifying documents on keywords, keywords could be
10:07
anything, the documents could be in the internet, tweeters or newsletters or something like this, emails and the keywords, you could give the keywords or you can extract them automatically for frequencies, this is just to make the more appealing
10:26
for the participants to know what that the data mining is much of looking about structuring the data, so you can end to find, but first of all talking about the data itself now,
10:40
just to have an introduction, I will not take much time about it, it was just to introduce that this is all classified, data comes in tables, there are several rules or labels for each column, normally there's a very sweet column, you call the label or the target that you want to predict, you know the other columns are the rubular that over the labels which we will use
11:06
to do the prediction and this was just in the way because I was trying to, this will be to introduce because in Python you have other names for this and there are several names
11:22
and in trying to say that we have tables but we can have a multi-dimensional arrays with tables, normally this kind of 3D arrays is a lot used to get image sets, for instance you have image compare even compressed in a data set, so you have several image like it was like a slice
11:41
and then the problems with that, it was just it's normally this all of these kinds of problems with the pre-processing phase that takes almost, I don't know, 80% of the time is to run all the data because they can come from several places, they have to be
12:03
shaped and cleaned and so on, so you have to deal with a lot of the missing data, duplicate, incorporate, encode, there's no other types, reshape that to your proposals, dealing with outliers and even trying to reduce the dimension if you have too many of them
12:25
but first we have to deal with the explored data, so I was like to plan that after I introduce you the hard way then I say but you have to start with the exploratory data, you have to do the typical analysis of the descriptive statistics, no it's
12:43
we end with visual data, so you have to use like the univariate data into analysis where you are analyzing each of your columns to see about the shape of your data or you have to analyze it by pairs to compare it with one another, no association to calculate
13:06
association, mainly you can calculate association with your label, your target value and then this is a standard caution warning that all you are looking for correlation in things so
13:21
don't get too enthusiastic about the cause effect of things, that could be a third or four factors in the middle of your, so I just introduce you then to whether to whether models or methods for that I use more in my investigation, here I'm not
13:46
presenting it would be to use reducing the dimensionality by PCI and to do some clustering and with a simple one that is with weak means form and just trying to show them how you what
14:00
PCI is, how you is you can compare it as a best fit, introducing a best fit plane or shaping your data and then reprojected in that plane so you can produce a clear vision of your data, normally this is very used for plotting your data and instead of
14:24
plotting in the space of your features, you put it in the space of components that are artificial features produced by this method, it's a standard method and then you can even use it to clean up and to select attributes and clustering the two main approaches, there are others based
14:46
density, but the two main are something like a top-down approach with partitioning the space or the bottom-up where every example it's going to be joined in the cluster, that it's the main feature for associating data that you don't know
15:07
what are the, how they are associated, if there is any structure in the data that so you have to choose something, you choose distances, dissimilarities and try to
15:21
pick groups in the data automatically, this is an automatic thing, there are several clustering algorithms, this is all very visual but the data is abstract and some have some pros and cons and I explain a little bit more the k-means algorithm, very used and
15:46
and it's simple and normally used for big lattice sets because it's more fast also but it's got a lot of familiarities with Voronoi that divides the space in it's an interactive algorithm that goes finding the distance between the
16:06
centroid and the mean of the value, so it's centroid that is trying to to will always displace itself to the mean values of its elements, of its cluster
16:21
so this will look until a place where the interaction stops and there is no more movement you'd say like this, in the end just I'll just tell some words about overfitting and and the problems that you deal when you deal with that, normally if you have
16:43
very many dimensions, many attributes when you are analyzing how many attributes is space it's potentially the volume is potentially with the number of dimensions of the values so if you have a lot of values you would need to have more exponentially more examples to
17:05
fill the volume so it's not, sometimes the data gets too sparse and there are some combinatory expulsions and some BS and the main is the overfitting so you have to have a general, your model shouldn't model all your examples, you have to
17:24
accept some error in the way that you have a generalization so to explore we started with some visual programming, RapidMiner it's very used it's very important and I thought that maybe I had architecture students so they know where
17:44
are super and so on so this would be a nice way to introduce a visual algorithm composer that goes with the data flow from left to side to right and mainly I was introduced
18:02
that and because they had even a wizard you can do all data mining process in a wizard step by step with five steps it's powerful but even from collecting that and cleaning it, selecting the label and choosing the models they analyze
18:23
five or six models at the same time then you have a score and so on and then you can build the algorithm that was behind the wizard so you have a not a black box you can then fine-tune everything but show it also orange it's a very common it seems like for kids but it's very
18:47
powerful and it's all also visual and it's very interactive but you can deal with the images text and the rule induction and so on it's a nice environment but we were focused on Python and
19:05
it for me I am not an expert I should say it was an opportunity for me to explore more the Python environment as it becomes a very common language across across many scientific fields and especially in the analytics and data science but this was just
19:26
I have no no special support for Python but it's through that it's getting more used and I think it's a nice language for no architects or GIS people because you can
19:43
learn from you are applying it in the in the data analysis environment but you can use it in algorithmic design then it's it's very connected or GIS if you are built or even in map and I show these were the main libraries the libraries of Python to to deal with data
20:03
analysis from from scientific computing and crashing numbers to getting data from the internet and everything then dealing with data frames with the tabular data and then the ski learn
20:21
the ski learn with the with the machine learning and this is the environment normally there's a lot of the libraries and they are very interconnected it's so you have normally the best weights and there are several packages of Python that bring all the libraries necessary libraries together
20:41
so this is an account it's very useful data mining and data analysis and um the way is because they bring all the libraries and because configuring a standard Python it's not that easy it's a lot of libraries depending on libraries depending on libraries
21:03
so and this is the aspect of a we work with with a anaconda and use a specific way of coding with pythons the notebook the jupiter notebook it's a very well it's almost like a word document or if you want but with the cells and it shall and cells can have images can be
21:26
plots of the or live code inside so you can print it you can share it it's a nice way of interactively and to share your investigation you can go here and put your things uh and there are some special libraries for dealing with them there are libraries for everything but
21:49
there are a world of libraries for python but this for mainly for special analysis and and geographic information systems you can deal with shape shape files and so on and so
22:02
uh collect the internet from the open street maps and so on the examples we we did some simple examples like was not so this is like a importing a a csv from a from a from a vga analysis and plotting the discrete the summary this almost the
22:25
statistic summary of the the columns then try to make some the plot using the there's a very powerful ways of configuring plots any way you want it's so you have the map of the the plot of a vga
22:45
analysis with the with the histogram and so you can even put the color bars and you can configure or do more complex multi-dimensional analysis with this scatter plot table
23:00
here it was just an exercise showing a little bit more deep how this y scripting can help us for instance to calculate the number of the clusters in a with k-means this algorithm for clustering you have to give the number number of clusters and that that's a good thing it was
23:21
supposed to you don't know uh think anything about your data you can have a feel and experiment in scripting you can run in a batch way produce a lot of models with two three four five six seven eight and then to plot a lead plot is what is called the i have to read it is called
23:42
the explained variance and there is a neuristic something ethical when there is an elbow in that graph you should choose the group corresponding number of clusters there is in the bottom of that this is a neuristic to calculate the number of clusters based on 4k means other algorithms
24:01
calculated for you it's not that one this is for instance i didn't show this in the workshop i should have said better this is to connect for instance from open open street map directed to your notebook data from open map or in here with a black and white in the figure
24:22
ground figure like get collected and this is go to your to your notebook and you can keep on with the with the network analysis with the libraries of network analysis and so on so yeah there were the other part was to show the power of scripting so it was to so on power was to show that the scripting and in this case that even inside the
24:47
grasshopper that i thought that many people would use it in in this because they were architected they would find this is using clustering in of the geometric figures it's a simple example of clustering geometric figures with k means i think or here to classify digit
25:05
and write digits inside the inside of the grasshopper environment so you know and to plot it or do a regression from with a neural network which are there are components that try to
25:21
implement all these libraries that came from python inside the environment design environment so here we have the only the dots in this is fitting the the surface for that dots giving the sheets on the only the planning he finds the in a very simple way is a fitting
25:40
on the surface on it even you could do scripting there you could even do actually doing some scripting inside them inside the map because oh i thought or they are on the deep map side or they are in the algorithm design so there is a it was something that i knew and i want to explore so this is very recent and i never explored it before but there
26:06
is a a simple scripting language for deep map called solid script and you can do several things with it we normally use that window to to to create new fields with a formula or
26:22
something like this or and but you can start to put some code if then and some simple functions it's a very limited the it's not language scripting language with three or four functions and so but for is the the you can do and if then to select for instance if you are above
26:45
the mean the medium the mean of the of some property or or below and you can this was the part that i was looking for it so you can assess to the neighbors on the graph
27:00
deep map normally doesn't allow it because it processes all the all the points and they give you the final here you can assess more fine tuning the the elements you went in the graph that is beneath the that was created before no so you have a graph and you can go querying for these neighbors you can go where the superposition of two vis-a-vis or something like
27:26
this this is just examples and don't get a little bit more deeper that would be this would be plotting the minimum value of connectivity that the neighbor that you can see i'm here and i plot in my place the minimum value of connectivity and some structures appear i don't make any
27:44
theory about it but it seems like some path or something like this some path will will be opening because something i didn't really this is very recent but you could make some opening a little bit because you go from a narrow place together and so it seems that the path there
28:04
highlighted like this way or something like this or you can plot them for instance in the same way instead the minimum the maximum so do you have partitioning the space in the kind of vis-a-vis where you are mapping the maximum value of in the exovis all across the the system so you
28:24
then you can retrieve it so i don't know something like this would be the where are the maximum values of the i think it's connectivity in this case and um important no because the the maximum local values so you can go across the to it to see where they are
28:42
and to see just for testing yes if you make a visual step with the they are it covers all the system because you you are recollecting all the places or you can just isolate the site of interest and the and work with the making some kind of connectivity based
29:03
only on that not not on all the all the graph it's just to illustrate this you could isolate it and then you can use it it'll be something like or it will it is an error to like a rainbow is obvious or something like in the end just for uh for concluding marks
29:22
data mining it's used in the in the lot of data mining methods now they are used in design and mainly even in design and in design and analysis so more in the analysis but in design they can be used because so because they have a strong title optimization for this
29:41
but all of this you can are part of the edge became part of this kind of of analysis where you can start to get melting data from several places now mainly social net or social media that using data scribing to assess a more or less legal way of assessing
30:05
that of us the database in the internet where you create crawlers in the internet and you collect the scenes geocoding and to analyze unstructured data and management of database visualization and optimization and and that would be related with scenario creation so you can have
30:28
some optimization based on constraints so you can you can do it and nothing in the end it was this and just to say that this would be a this subject this would be a way i think to
30:44
usefulness would be to assess the open data for instance that is becoming more available there are a lot of big databases now that we can more public you know and another point that some people involve a point is that a way of involving the younger generation the digital
31:05
natives that born with internet this would be to now they have not met but they have information to to to meddle with their hands and they are all the time creating it and to get more insight of just going behind the scenes and how that how your that is being treated
31:24
know why in the internet there is an algorithm that can recognize our face so but how does it do it it's not nothing of the other world you can you can learn it and and use in design would be also to create richer programming for the for the design to have
31:43
more information and do its thing things that are more meaningful to the user because you can collect this this amounts of that but there's a lot of problems also because it's not that easy to to implement there is a black box effect many algorithms are black you don't know what
32:01
happens inside them normally things like neural networks and so on it's very difficult to understand so and and and it makes like a yeah i think it says some risk of making reality go
32:20
more far away because you have that in the middle and as a potential of the as a as a pointing it as a problem is getting the reality more apart from the designer or the or the or people in general because you have such a thick layer of of information in the middle
32:40
and in this thank you thank you for it thank you i'm calling now patricia also she's a giant professor of the department of
33:00
architecture and urbanism phd student in architecture and urbanism from the post graduate program in architecture and urbanism in uftb currently doing part of our phd studies at the spatial morphology group at schalmers in sweden
33:26
well the presentation will be on development of permeability measure between private and public space good afternoon my name is patricia lonzo and my presentation is the development
33:44
of a permeability measure between private and public space and the authors are myself from brazil from chalmers sweden and louis from also brazil
34:09
and the content of this paper is part of my phd studies on the relation between density urban
34:21
form and urbanity and so the natural movement theory by hillier says that the street configuration is the primary generator of pedestrian movement and activities and many authors agree that high density generates a higher intensity of activities and movement and cities
34:49
however in certain urban contests like latin american cities and in the specific study case of this paper brazilian cities this may not be the case in that contest sometimes
35:09
uh high street centrality and high density they come with low presence of people in public space and our hypothesis is that this is caused by the specific way densification
35:28
takes place the present-day brazilian cities they have a typical verticalization process mailing for housing with high-rise residential buildings as you see in these photos these
35:42
are in the study case city of this research and these buildings they have semi-underground floor and often also a ground floor for car parking because of the the mobility
36:02
policy that prioritize the private transport so here in this paper we analyze two characteristics related to this densification type which seem to contribute to less inviting and less
36:25
safe streets first the bigger size of plots these higher buildings they demand larger plots because of urban law requirements and for that several plots they are joined as you can see
36:42
in the pictures reducing the number of plots per block and second the consequent lower frontage permeability there are fewer entrance and blind walls due to these semi-underground floors for parking and so there is less interaction between buildings and streets so we needed a more
37:09
comprehensive and thorough frontage assessment and then we developed a measure that captures the frontage permeability both in a qualitative and quantitative term by verifying the frontage
37:27
visibility and accessibility separately the presence of setback its depth and use and the type of space referring here to the land use which there is permeability too
37:47
so we define the frontage permeability as its property to allow interpersonal interactions in both visual and physical ways between the public and the private space delimited by them
38:04
the study looks at the public private interface from the public space perspective and the permeability measure as I already said is divided into two variables that are measured
38:20
separately the visibility which is the property of the frontage to allow people to see through it and the accessibility its property to allow people to pass through it and the measures they come they combine the degree a quantitative measure of how visually or physically
38:44
permeable the frontage is to a qualitative evaluation about the type I mean the land use of space there is visibility or access to so in this table we can see the permeability is divided in visibility accessibility and both of them
39:04
they have a four category scale so for visibility we have no visibility and then visibility to empty space and that means vacant plots or residual or underused
39:21
space and then visibility to private space and finally to public space and then it is seen that the setback if there is or no setback and for accessibility is the same no accessibility no accessibility sorry no access and the
39:49
control access to private space and that is when you have a gate or a door and this private space they are residential use the control access to public space
40:07
all other public semi-public or collective use like retail service etc and finally the open access and again it's we check if there is setback or no
40:25
the entrance of closed or abandoned buildings and entrance of garage they are considered of as no access because it is understood that there is no social interaction in this case
40:43
so the data gathering was made by google street and google earth the mapping with autocad the data processing and analysis with gis software and the visibility values were based on the average height of the eyes of a pedestrian walking on the sidewalk level
41:05
and the high limit to open access which is a barrier with 40 centimeters high considers the intention of territory demarcation of course you can pass through it but we understood that accessibility is not only about being able to pass but also about
41:27
permission to pass the lengths were measured in meters and each plot frontage is the sum of the several entities according to the variables values for visibility and accessibility
41:44
the geometric representation is with lines with attributes according to that table that i showed you and the line length is also added like as an attribute and in case of a setback
42:01
its death and its use so the length of each permeability category was measured to then develop various indexes to describe the frontage permeability of each plot so this is the calculation of the main indexes the visibility one and accessibility one so they are based
42:26
in the total length of the plot frontage and we also developed some sub indexes that we found that should be important like the visibility and accessibility index to public space and the
42:45
visibility and accessibility indexes to both public and private space and another one is the accessibility index for open access the pilot study
43:06
are selected areas of manaira district in jon peso a city brazil these photos are from this district and it is an economic functional centrality because it is a subcenter
43:24
in the the city with housing retail and services it is also a demographic centrality because it has high population density and also a morphological centrality because of its high values of density and integration choice i mean the syntax measures and accessible density
43:50
this accessible density is measured with the place space tool it's a plugin for qg's that use syntax analysis for measuring the accessibility through the street to different
44:02
contents of urban space for example density that's one the one i use so based on the natural movement theory again we would expect high levels of people movement in this area however in the preliminary observation this is not it it does not seem to be the case
44:33
and here we have a map with the four areas the testing areas the areas one and two i don't know
44:43
if you can see um they are in the north part they have the highest values of accessible density in local radio within the district range and the areas three and four
45:01
they have medium to high values in the same region and they are next to to malls that act as attractors so as the testing areas have high centrality and accessible density uh in a district that is also a centrality it was supposed to to be expected uh urban vitality
45:30
but according to our hypothesis there are other variables and they are the plot size and the frontage permeability that may affect this relation between density centrality and urban life
45:46
and in order to test this apart this hypothesis the permeability measures are related to local density and i mean the local to to to make a difference from the accessible density that is
46:03
a configurational measure so the local density it's basically fsi the floor space index plot size frontage length and land use this paper does not relate the morphological data
46:24
to people presence in the streets this is the next step in the research and here we have the maps of the four testing areas showing the frontage accessibility it's not a good image also
46:42
and here the frontage visibility in all areas there are many frontage with low visibility and the findings with statistics we cannot say that there is a statistical correlation between visibility and accessibility indexes to the density values but we found the pattern
47:08
in the low and medium densities up to fsi3 the entire range of visibility and index and accessibility index is present the entire range of it we can see there
47:25
at high densities fsi above three to eight both indexes remain low so this pattern is repeated in the indexes comparisons with plot area and frontage length
47:42
it also occurs with the visibility and accessibility sub-indexes for public space and for both private and public space and this repeated pattern confirms the hypothesis that this high-rise residential buildings type
48:05
cause not only urban densification but also a decrease in frontage permeability and there is no statistical correlation between accessibility index and visibility index that
48:25
means that these variables do not necessarily come together which confirms the pertinence of the proposed method in measuring these two variables separately and we can see also in this
48:41
chart that both public use plots retail and service and those with all other use present high and low frontage permeabilities also in plots with other use the low values predominate this indicates that the frontage of public use plots are not necessarily more
49:06
permeable which is confirmed in these two charts which relate the number of plots with public use and other use to the indexes the visibility one and the accessibility one
49:25
so in yellow you have the public use and blue the other use and there is a strong correlation between the indexes of accessibility and visibility to public space
49:47
which indicates that when there is higher visibility for public space there is also higher accessibility but in the indexes of accessibility and visibility to both private and
50:05
public space they when they are together they have no strong correlations so when we analyze this together we can say that in the frontage of residential space there is a mismatch between accessibility and visibility the sub-index accessibility for open access is always
50:30
very low i didn't show it it's very low in any graph so it demonstrates that open access is it's not it's very uncommon in this area and the setbacks are mostly used for parking
50:46
so they don't contribute to social interactions and the moderate correlation between density and area and between density and frontage length showed that there are still many buildings with
51:02
high fsi i mean density in relatively small or medium plots and with low or medium frontage length so the hypothesis that the verticalization type in focus brings about bigger size plots is not confirmed and the conclusion first measuring the frontage visibility and accessibility
51:29
separated has proved to be pertinent because these variables do not necessarily go together and verifying the land use is also relevant because the frontages of public space are not
51:43
always more permeable but when they are they present higher visibility and also higher accessibility which does not necessarily happen in private space frontage so the initial result sorry the the initial results confirm the hypothesis that the increase
52:05
intensification model in brazilian cities with high rise residential buildings generates a decrease in the frontage permeability although it does not change the plot size
52:21
and the frontage length in a significant way and and we can say that the dissemination of this building type generates low interaction between streets and plots since it's often surrounded by blind walls with no visibility and few accesses especially when they are built
52:42
side by side according to the results there are also less dense types that have low frontage permeability and this is about our urban problems the fear of urban violence it's
53:04
it's very present in our cities however this case commonly have isolated walls which can be easily modified or demolished to generate higher permeability but in high-rise buildings most of the frontage are semi-underground garage walls as you can see in these photos
53:27
and they are more permanent structures and more difficult to modify so this is a bigger problem and it's important to stress the role of urban planning and management in controlling this
53:43
lack of frontage permeability the urban law should define minimum percentage of frontage permeability and stimulate the mix of use and i am finishing i want to say that these are preliminary results this is not big data it's an ongoing investigation other empirical
54:11
studies need to be carried out for more complete and representative data and the next steps of the research are expanding the testing areas and relating the morphological data
54:24
to people presence in the streets thank you very much obrigado well last but not least
54:45
i will go with the kind of the sewer well she's a PhD professor in the university of London of Lisbon sorry
55:04
and the main topic of work is on learning spaces a presentation will be on the centrality of vocation-oriented knowledge and assessment
55:21
of the location of polytechnic institutes in Portugal hello everyone i'll try to to be brief and to deliver a light presentation since we are finishing this wonderful symposium as franklin said i'm with the kinesis and i work with tres etor and mofalo toucan at situa the center for innovation in territory urbanism
55:45
in architecture at technico in lisbon and i'm feeling a little bit off topic here because my focus is really on the case study because even though we do use formal methods and space syntax
56:00
concretely we focus on learning spaces in in my research group so we've been working with higher education learning spaces for the past years and in this case i'm focusing on polytechnic institutions in Portugal because higher education in Portugal well worldwide
56:22
has changed a lot and a lot of pressure has been imposed on on higher education infrastructure and of course we have to adapt to so many things as the massification of higher education the global changes and for instance the implementation of the bologna process which
56:42
was signed about 20 years ago and still it feels like our buildings are not able to react to all these changes so we had a highly changed higher education system worldwide and then we have buildings and infrastructure that are not adjusting at the right pace so
57:05
of course we want to analyze these infrastructures and to understand how can we improve them and how can we make better decisions in them and how are they related to better missions or to different missions of higher education mostly focusing on social aspects and
57:23
interaction so if we have a higher education system with a threefold mission based on teaching research and outreach in Portugal we have to differentiate between universities and polytechnics and even though recent OECD studies have shown that this tendency in some aspects is getting
57:45
is getting blurred still one of the main assets of Portuguese higher education system is in fact this binary structure and polytechnics have been perceived as very successful infrastructures in combining and attending to vocational knowledge so according to Portuguese
58:03
law they are oriented towards the creation diffusion and transmission of professional type culture and knowledge and they should focus on vocational training and professionally oriented technical development considering this then we have to look at the geographical dispersion
58:22
of knowledge institutions in Portugal and on your left hand side you have the map of in universities in Portugal and we can see a clear tendency towards the location of of these institutions in the main cities Lisbon and Porto and in the coastal area but then we have polytechnics spread more or less all over the place and of course they have
58:44
a huge responsibility in educating the the more segregated areas because these places can really be connected to different kinds of population and be a key asset in developing the country in this case we are focusing on two of them especially because we are applying a methodology
59:04
that analyzes the institutions starting from the city so we consider the city at its global scale and then we go deeper into analyzing the institution in order to understand how do these institutions respond to their mission in terms of being open to society open to the
59:22
academic community and providing spaces that can be used by the public in general so we are working with Bregense in the north and Beja in the south because these are the extremes in terms of polytechnic institutes also because they have very similar conditions so
59:42
the cities are more or less the same size Bregense is two square kilometers larger than Beja and they have roughly 25 000 inhabitants each so they are very similar in demographic terms if we move forward
01:00:00
to the axial and segment analysis, then we can see that their structures are in fact different. So while Beige has a deformed grid system that is grounded on a radial structure, Bregenza is much more organic, much more relying on topography and with the strong influence
01:00:21
of a valley that crosses roughly north-south. We can also see the fragmentation in the structure if we analyse the numbers on the axial and segment maps because we have a ratio of relationship between axial and segment lines that's smaller in Bregenza, which is consistent with its shorter lines, and that
01:00:45
reinforces the levels of fragmentation that we can perceive. So when we move to the axial analysis and we try to understand the position of the institutions within the structure of the system, the first thing that we can see is
01:01:02
that they tend not to be very integrated. So in Beige, the Polytechnic Institute is located to the west of the city, I'm not sure if you can see it from there, and in Bregenza, roughly in the geographical centre, a little bit south. But the cities present very different structures in terms of syntactic properties.
01:01:25
So the integration levels in Beige are higher than in Bregenza. So the structure of the city is manifesting here. And also the Polytechnic has a better average value in Beige than in Bregenza.
01:01:41
If we analyse its main axis, because we think it's a key issue for visibility of the universities, the location of the main axis, again we are still analysing integration and in Beige there is a better response or a better potential for the university to
01:02:01
become a destination. Still not very good because in these cases both precincts belong to the 50% more integrated areas of the city, which is not that good. When we move on to analyse betweenness and centrality, then the results are slightly more encouraging.
01:02:21
And we can see that in both cases the precinct of the Polytechnic is sort of relying on the foreground structure of the city, even though composed mostly by background structure. So it means it behaves more or less like a neighbourhood, but grounded on the most prone to be used path in these locations within the city.
01:02:47
Here although we can find much better values, so the cities perform better in general and they even fit into the normally considered like the reasonable values defined by Hillier
01:03:01
for this variable, the angular choice. When we get to analysing the main entrance, then the results are way more encouraging. And this is in fact consistent with some analysis we've done previously with universities worldwide that shows that even though the universities tend to be somehow segregated and private
01:03:23
within their precincts, they usually have some more visibility in their main access because that makes them attractors and that makes them part of the urban landscape of the city. So they are visible and they get a physical presence within the urban tissue.
01:03:44
Then we went a little bit deeper, so we've analysed the precincts and we chose two specific buildings of each one. We've analysed both the agrarian schools of both cases and also the health schools of both cases. And while the agrarian schools are located in campus, the health schools are located
01:04:03
off campus in Bregense because they are connected to the health centre in the area. In Beja, just by proximity reasons, I think they didn't have any more plot available in campus, so the health school was developed a little bit more recently and it was developed
01:04:20
outside. And we started by creating a functional analysis that separated the sorts of spaces that we wanted to analyse. So regarding the elective spaces and academic spaces, there is not much differentiation, but we've analysed specifically the places that could be used by the exterior community,
01:04:40
so auditoriums, cafes, libraries, information centres and so on. What we found was that the structure of the building somehow responded to the structure of the city and we have a more fragmented and difficult to read structure in Bregense than in Beja.
01:05:01
Also the topography doesn't help and here there is something that might be of importance because in Bregense the schools tend to have more connections to the exterior which of course seems like a good thing when you want to reach the auditorium or the bar or anything. But then you have the power of the keymaster so we actually don't know in situ how does
01:05:22
this work because this structure can be severely affected by the man that locks the door. More or less the same pattern for the health schools so we have very deep schools if we consider that they reach seven levels of deepness so if a student wants to go to
01:05:42
a classroom on the last level it's still a lot of effort to reach. This is just another organisation to simplify things and it shows that there is at least one very good issue is that we have a complete mix of uses so we cannot say that we have classrooms on the easiest level and then research areas in the top one it doesn't happen
01:06:04
like this and we have a complete mix of uses and at all levels it's possible to find almost all sorts of spaces. So to conclude we don't have a lot of conclusions because this is work in progress and well of course I only shown two cases we have some more but still not enough to
01:06:24
to have support all of our information and our research so I would say the most interesting thing is that really universities don't seem to have changed according to the changes
01:06:43
in missions that they have had in the last decades and actually we are still building campuses as we did in the 19th and 20th century in America so we really should start thinking about how do we want to build for what we want to teach and all the student centre
01:07:01
processes that we have and also some of the types of learning that we've seen here in this symposium because most of our spaces are really not responding not even to the social changes that we have so it's important to consider how do we want to teach our students in the future and how do we want our spaces to convey the missions that we are supporting
01:07:22
and that we are spreading. So thank you for bearing with me in the last part of the symposium I appreciate that you are still all here. There is a thing I would like to, oh please please.
01:07:40
Why some places there are a lot of people and the other ones there are not much and I was thinking like you decided to do this research as describing how the spaces are but what if you try to look for how many places to go there or how many places can people come from.
01:08:03
Wouldn't this give you a better reading of how many people, vitality of the street than just describing if you can or not go there. We passed, the way here I passed through a few squares that are very beautiful and completely accessible but it doesn't mean that I'll go to that place to check this vitality that
01:08:24
we are searching to describe right. How many people been on the streets was pretty much what you are trying to describe and see if that space could define it. Yeah, what will be very important to me and I don't know the people movement, the data
01:08:51
about the people movement in the streets so I intend to go there and to measure it and to try to correlate with this morphological data.
01:09:03
But to see where are they going, well it could be a way but what I wanted to check was the relation between the density and the form of the city and the social process
01:09:25
that can occur in the city. So when I go to the literature we find a lot of authors that define that high density
01:09:45
likely will bring more people and the syntax also tries to show that more centrality also is important to co-presence so I try to find an area with high density, with high centrality
01:10:08
but there are other varieties that is the frontage and the plot size and I think
01:10:20
that these characters they come with the type of the building and they can also interfere of course they are not decided but they interfere somehow maybe in this presence
01:10:41
or not presence of people in the street. As it is central and as it is dense it is supposed to have people there because it's a centrality, it has housing and it also has retail so okay I can try to see
01:11:04
where are people going but there are people living there and these people they have to leave their house so what I am trying to say is that
01:11:20
Because there is a bunch of people there, there should be many people in the street There is a lot of things that interfere in this, for example the urban problems in Brazil people fear the street so it's very common even the houses that are not so dense
01:11:46
they have walls but what I was trying to find is that with more densification with these high buildings this lack of permeability is it getting worse
01:12:04
and my data, it's still a little data but it already shows that the more dense we have more lack of permeability and of course it interferes in the way people feel in the street
01:12:24
I don't know if I was clearing Okay, I'll change the question More questions? Patricia, I have a quick question Do you think you are going to research the variations of typologies according to your variables
01:12:46
say certain types of building make the situation worse for co-presence and certain improve it or I mean within this range if you are going to try to find sort of differentiation
01:13:03
Yes, and the first thing is the density When I have higher density usually it comes with verticalization It's a process that occurs there and the densification is by verticalization
01:13:26
and this kind of verticalization with these high buildings for houses they have a very clear type of this base that makes this wall
01:13:50
and when you have not so high buildings it's another type that sometimes is more permeable
01:14:07
but the thing is that in all areas it's very common to find lack of permeability That's why the graphics are not so clear because there is no correlation
01:14:24
because when you go to low densities they also show lack of permeability but not always so it's spread but when you go to the high densities it's always low density
01:14:41
so it's very clear that the type is showing this relation I don't want to say correlation in a statistic way, it would be wrong but they are connected
01:15:11
I think the study is very interesting because it shows how different places in the world
01:15:21
because of their different setting and their different organization a measure that is developed at one place may not necessarily apply to another and looking to the situation in Singapore I think your study seems quite relevant because we also have a lot of residential
01:15:47
I mean most housing is high rise or relatively high rise but there are two very distinct types one is public housing which does not have any wall around it
01:16:03
which is completely open and even has the ground floor completely open through void decks so you don't only have the permeability but you also have porosity and then the private housing which also has all these walls around it
01:16:26
and I mean of course there are still differences but just reflecting on that it's quite obvious of course how one has a lot more potential
01:16:40
and you also see a lot more people walking around than in the other situation Something Patricia said touched my heart Three words Ippotis is not confirmed I know Jean Ventour talked a lot about his work
01:17:04
and he says in our discussions that he has not in his work of several years he has not arrived to great conclusions Well this modesty and dishonesty is absolutely marvelous in our work
01:17:25
in all your presentations there are honesty and modesty and this is really marvelous We can compare with some other speeches on architecture
01:17:42
when there is a clear sentence that sets the truth and there are norms, there are normativity and I have nothing against normativity but when normativity tries to be the truth
01:18:03
that I cannot allow I think this approach we have the use of formal methods is one of the ways we have to affirm this modesty and dishonesty
01:18:25
Because it's very difficult to deny the facts We collect the methods that are explicit
01:18:41
They cannot be falsified Well I think it was this I would like to say Thank you, we are ending now and we are going to...
01:19:04
Well, to where? I will close it Okay Thank you very much Only a few words to finish
01:19:21
Dear friends, the FORCE International Symposium on formal methods in architecture will soon be coming to an end On behalf of the organization I should like to express our most sincere gratitude to all of you for your combined efforts It has been a great honor for us to spend these past four days with you
01:19:40
With your activity participation the symposium successfully completed all of its goals This was the most rewarding thing to witness The primary goal of the symposium was to bring together researchers from all over the world who would be able to engage in a likely and intellectually open dialogue discussing the issues facing architectural formal methods
01:20:02
making constructive contributions towards strengthening our research practices and approaches helping to promote interdisciplinary exchanges and to explore new research questions We decided to begin with a model adopted in previous formal methods symposium and use this to create a new and exciting meeting combining contents of quality
01:20:24
and adding some stimulating activities such as workshops or social happenings We also discard the possibility of parallel sessions to ensure the cross-fertilization of the various scientific fields We knew it will reduce the number of possible papers accepted and the quality of communications presented
01:20:43
but the sponsors agree with it and support this decision After all, we don't seek quantity but quality so we are totally convinced we took the right option Obviously, we can improve the way we did things and get better in the future
01:21:01
Apart from the keynote sessions, the most powerful impression that we have formed over these days has been related to the quality of the papers presented and the debates that gave rise to providing us with the opportunity to scrutinize both the state of the art and the state of the practice in the architectural formal methods umbrella field
01:21:23
to present advanced experiences, exchanges of ideas and major practices and propose many concentrations of ideas and suggestions for improving architectural formal methods from a theoretical and methodological point of view to discuss some of the challenges facing architectural formal methods today So, AFM
01:21:43
Some concrete ideas emerged and many new connections we made that we are sure will have a positive impact in this field We believe that the closing of this symposium will not be an end but a new starting point What comes next?
01:22:02
We can maintain professional or and personal contact between us We can invite each other for future meetings, lessons, workshops or combine and merge expertise to provide mutual and complementary service to society We can even build a platform for international cooperation in SCABA
01:22:21
This is a new brand, a new acronym that we invented Space Configuration, STBT and Visibility Analysis that cover a large spectrum of our work So, we built a multipolar network to communicate between us and with the society to discuss and support people and organizations who are designing the future of the cities and the environment
01:22:45
The OPPO Art Project team is now implementing a close at Porto that can be linked to others around the world Before I formally close the symposium please allow me on behalf of the organizing committee to express again our sincere gratitude for the support of the OPPO Art Formal Metas project team
01:23:04
in making it possible for us to host this symposium and for the participation of all those involved I would like to thank the institutions that support this symposium The SESAP, Cooperativa Dincinio Superioristico Duport, CEREL to ESAP, Scholos Superioristico Duport
01:23:23
and its master program in architecture for the institution of sport The North Wind Wind program and OPPO Art Formal Metas project team which allow to fit in our activities the support of the European regional development found from the European Union I should also like to extend a further word of thanks to the various people who helped to make this symposium a success
01:23:46
I make an account and I suppose more than 300 people are involved in this Not only the people that are here, but all the people that committed something to this symposium So I should like to extend it to the group of keynote speakers, of course
01:24:04
It was an honor for us all to be able to listen and to discuss your ideas We link it remotely to real generics also I would like to the group of volunteer chairs who conduct the paper sessions Unfortunately you have the gong power, you and me have the gong power
01:24:22
to shut down discussions due to the time constraints As you know in Portugal we are ferocious with delays We cannot support delays at all I would like to thank the dedicated staff of the meeting desk and the support team I just want to let you know, the support team and staff here
01:24:44
that having you in your team made all the difference Next time I will make t-shirts and give you a t-shirt that says I am outstanding beautiful To the members of the scientific commission for their effort, availability and commitment
01:25:01
They will continue to work on proofreading articles for publication Of course you will work a little more Work has not stopped yet To all my colleagues of the organizing committee with special thanks for Professor Franklin Muraaj, Katerina Ruivu, Ixin Pashau without whom it will be impossible to carry out this task
01:25:22
Also to David Lait, Vienna This is a reaction for their continuous effort and contacts with several institutions and personalities David can fill my mail with an average of 135 mails per day To have been with me working at the forefront over the last years
01:25:43
I know things have been a mess sometimes and the rush hours were tough But you did a great job I know Professor Franklin's list of tasks to perform are increasing week by week Nowadays lists have more items to do AACP than in the beginning
01:26:01
And the symposium is at the hands So I will need more 9 months to complete it all To all of you for your passions with all the various inconvenience that may have been caused by our less than efficient organization of the event And finally I will thank of course because we try to think about everything
01:26:23
But we give the best but sometimes the things don't occur well So an event such as the symposium which brings together people from so many different parts of the world demands a great deal of organization, planning and time consumption I feel, as I am sure the others do, that these efforts have been rewarded
01:26:45
by some of the most creative and thought-provoking sessions we have held so far And everyone connected with the academic formal metallurgies are very grateful for that I know of course that symposium happy hours are shortly
01:27:00
I know that, sorry, must be I hope after all this four international symposium formal metas in Architecture held at Porto in 2018 will leave a memorable and lasting impression of all of you Now it's time to mingle once more I hope to see you in the conference dinner
01:27:24
Thank you so much for making this sport symposium possible and for making it so truly memorable This four international symposium formal metas in Architecture is now formally closed Thank you